From 6e29ee4062908ba8e6124bf8b3a2083e1b2d3e36 Mon Sep 17 00:00:00 2001 From: Matthieu Herrb Date: Sat, 1 Oct 2016 10:17:45 +0000 Subject: Update to pixman 0.34.0. --- lib/pixman/config.h.in | 3 + lib/pixman/configure | 114 +- lib/pixman/configure.ac | 42 +- lib/pixman/pixman/Makefile.am | 2 + lib/pixman/pixman/Makefile.in | 8 +- lib/pixman/pixman/pixman-arm-asm.h | 37 + lib/pixman/pixman/pixman-arm-common.h | 11 +- lib/pixman/pixman/pixman-arm-neon-asm-bilinear.S | 12 +- lib/pixman/pixman/pixman-arm-neon-asm.S | 12 +- lib/pixman/pixman/pixman-arm-neon-asm.h | 20 +- lib/pixman/pixman/pixman-arm-neon.c | 24 +- lib/pixman/pixman/pixman-arm-simd-asm-scaled.S | 11 +- lib/pixman/pixman/pixman-arm-simd-asm.S | 566 +++++ lib/pixman/pixman/pixman-arm-simd-asm.h | 116 +- lib/pixman/pixman/pixman-arm-simd.c | 50 +- lib/pixman/pixman/pixman-combine-float.c | 338 ++- lib/pixman/pixman/pixman-combine32.c | 1686 ++----------- lib/pixman/pixman/pixman-fast-path.c | 2 + lib/pixman/pixman/pixman-general.c | 31 +- lib/pixman/pixman/pixman-gradient-walker.c | 2 +- lib/pixman/pixman/pixman-implementation.c | 16 + lib/pixman/pixman/pixman-inlines.h | 3 +- lib/pixman/pixman/pixman-mips-dspr2-asm.S | 2 +- lib/pixman/pixman/pixman-mips-dspr2-asm.h | 4 +- lib/pixman/pixman/pixman-mips-dspr2.c | 10 +- lib/pixman/pixman/pixman-mips-dspr2.h | 8 +- lib/pixman/pixman/pixman-mmx.c | 126 +- lib/pixman/pixman/pixman-private.h | 6 +- lib/pixman/pixman/pixman-sse2.c | 24 +- lib/pixman/pixman/pixman-vmx.c | 1229 +++++++++- lib/pixman/pixman/pixman.c | 35 +- lib/pixman/test/Makefile.in | 187 +- lib/pixman/test/Makefile.sources | 62 +- lib/pixman/test/affine-bench.c | 448 ++++ lib/pixman/test/blitters-test.c | 20 +- lib/pixman/test/check-formats.c | 176 -- lib/pixman/test/composite.c | 11 - lib/pixman/test/cover-test.c | 449 ++++ lib/pixman/test/fence-image-self-test.c | 239 ++ lib/pixman/test/lowlevel-blt-bench.c | 513 +++- lib/pixman/test/pixel-test.c | 2780 +++++++++++++++++++++- lib/pixman/test/radial-invalid.c | 54 + lib/pixman/test/scaling-test.c | 66 +- lib/pixman/test/solid-test.c | 353 +++ lib/pixman/test/thread-test.c | 29 +- lib/pixman/test/tolerance-test.c | 360 +++ lib/pixman/test/utils.c | 786 ++++-- lib/pixman/test/utils.h | 34 + 48 files changed, 8627 insertions(+), 2490 deletions(-) create mode 100644 lib/pixman/pixman/pixman-arm-asm.h create mode 100644 lib/pixman/test/affine-bench.c create mode 100644 lib/pixman/test/cover-test.c create mode 100644 lib/pixman/test/fence-image-self-test.c create mode 100644 lib/pixman/test/radial-invalid.c create mode 100644 lib/pixman/test/solid-test.c create mode 100644 lib/pixman/test/tolerance-test.c (limited to 'lib/pixman') diff --git a/lib/pixman/config.h.in b/lib/pixman/config.h.in index 17d825034..ab705701e 100644 --- a/lib/pixman/config.h.in +++ b/lib/pixman/config.h.in @@ -12,6 +12,9 @@ /* Define to 1 if you have the header file. */ #undef HAVE_DLFCN_H +/* Whether we have FE_DIVBYZERO */ +#undef HAVE_FEDIVBYZERO + /* Whether we have feenableexcept() */ #undef HAVE_FEENABLEEXCEPT diff --git a/lib/pixman/configure b/lib/pixman/configure index d6f22cbe9..a79b6206b 100644 --- a/lib/pixman/configure +++ b/lib/pixman/configure @@ -1,6 +1,6 @@ #! /bin/sh # Guess values for system-dependent variables and create Makefiles. -# Generated by GNU Autoconf 2.69 for pixman 0.32.8. +# Generated by GNU Autoconf 2.69 for pixman 0.34.0. # # Report bugs to . # @@ -590,8 +590,8 @@ MAKEFLAGS= # Identity of this package. PACKAGE_NAME='pixman' PACKAGE_TARNAME='pixman' -PACKAGE_VERSION='0.32.8' -PACKAGE_STRING='pixman 0.32.8' +PACKAGE_VERSION='0.34.0' +PACKAGE_STRING='pixman 0.34.0' PACKAGE_BUGREPORT='pixman@lists.freedesktop.org' PACKAGE_URL='' @@ -1390,7 +1390,7 @@ if test "$ac_init_help" = "long"; then # Omit some internal or obsolete options to make the list less imposing. # This message is too long to be a string in the A/UX 3.1 sh. cat <<_ACEOF -\`configure' configures pixman 0.32.8 to adapt to many kinds of systems. +\`configure' configures pixman 0.34.0 to adapt to many kinds of systems. Usage: $0 [OPTION]... [VAR=VALUE]... @@ -1460,7 +1460,7 @@ fi if test -n "$ac_init_help"; then case $ac_init_help in - short | recursive ) echo "Configuration of pixman 0.32.8:";; + short | recursive ) echo "Configuration of pixman 0.34.0:";; esac cat <<\_ACEOF @@ -1595,7 +1595,7 @@ fi test -n "$ac_init_help" && exit $ac_status if $ac_init_version; then cat <<\_ACEOF -pixman configure 0.32.8 +pixman configure 0.34.0 generated by GNU Autoconf 2.69 Copyright (C) 2012 Free Software Foundation, Inc. @@ -2193,7 +2193,7 @@ cat >config.log <<_ACEOF This file contains any messages produced by compilers while running configure, to aid debugging if configure makes a mistake. -It was created by pixman $as_me 0.32.8, which was +It was created by pixman $as_me 0.34.0, which was generated by GNU Autoconf 2.69. Invocation command line was $ $0 $@ @@ -3017,7 +3017,7 @@ fi # Define the identity of the package. PACKAGE='pixman' - VERSION='0.32.8' + VERSION='0.34.0' cat >>confdefs.h <<_ACEOF @@ -12201,13 +12201,13 @@ fi -LT_VERSION_INFO="32:8:32" +LT_VERSION_INFO="34:0:34" PIXMAN_VERSION_MAJOR=0 -PIXMAN_VERSION_MINOR=32 +PIXMAN_VERSION_MINOR=34 -PIXMAN_VERSION_MICRO=8 +PIXMAN_VERSION_MICRO=0 @@ -12308,6 +12308,53 @@ rm -f core conftest.err conftest.$ac_objext \ { $as_echo "$as_me:${as_lineno-$LINENO}: result: $_yesno" >&5 $as_echo "$_yesno" >&6; } + { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the compiler supports -Wno-unused-local-typedefs" >&5 +$as_echo_n "checking whether the compiler supports -Wno-unused-local-typedefs... " >&6; } + save_CFLAGS="$CFLAGS" + save_LDFLAGS="$LDFLAGS" + save_LIBS="$LIBS" + CFLAGS="" + LDFLAGS="" + LIBS="" + CFLAGS="$WERROR -Wno-unused-local-typedefs" + CFLAGS="$save_CFLAGS $CFLAGS" + LDFLAGS="$save_LDFLAGS $LDFLAGS" + LIBS="$save_LIBS $LIBS" + cat confdefs.h - <<_ACEOF >conftest.$ac_ext +/* end confdefs.h. */ + + int main(int c, char **v) { (void)c; (void)v; return 0; } + +_ACEOF +if ac_fn_c_try_link "$LINENO"; then : + pixman_cc_stderr=`test -f conftest.err && cat conftest.err` + pixman_cc_flag=yes +else + pixman_cc_stderr=`test -f conftest.err && cat conftest.err` + pixman_cc_flag=no +fi +rm -f core conftest.err conftest.$ac_objext \ + conftest$ac_exeext conftest.$ac_ext + + if test "x$pixman_cc_stderr" != "x"; then + pixman_cc_flag=no + fi + + if test "x$pixman_cc_flag" = "xyes"; then + _yesno=yes + else + _yesno=no + fi + CFLAGS="$save_CFLAGS" + LDFLAGS="$save_LDFLAGS" + LIBS="$save_LIBS" + + if test "x$_yesno" = xyes; then + CFLAGS="$CFLAGS -Wno-unused-local-typedefs" + fi + { $as_echo "$as_me:${as_lineno-$LINENO}: result: $_yesno" >&5 +$as_echo "$_yesno" >&6; } + { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the compiler supports -fno-strict-aliasing" >&5 $as_echo_n "checking whether the compiler supports -fno-strict-aliasing... " >&6; } save_CFLAGS="$CFLAGS" @@ -12718,15 +12765,27 @@ cat confdefs.h - <<_ACEOF >conftest.$ac_ext #error "Need GCC >= 3.4 for MMX intrinsics" #endif #include +#include + +/* Check support for block expressions */ +#define _mm_shuffle_pi16(A, N) \ + ({ \ + __m64 ret; \ + \ + /* Some versions of clang will choke on K */ \ + asm ("pshufw %2, %1, %0\n\t" \ + : "=y" (ret) \ + : "y" (A), "K" ((const int8_t)N) \ + ); \ + \ + ret; \ + }) + int main () { __m64 v = _mm_cvtsi32_si64 (1); __m64 w; - /* Some versions of clang will choke on K */ - asm ("pshufw %2, %1, %0\n\t" - : "=y" (w) - : "y" (v), "K" (5) - ); + w = _mm_shuffle_pi16(v, 5); /* Some versions of clang will choke on this */ asm ("pmulhuw %1, %0\n\t" @@ -12807,10 +12866,11 @@ cat confdefs.h - <<_ACEOF >conftest.$ac_ext #include #include #include +int param; int main () { - __m128i a = _mm_set1_epi32 (0), b = _mm_set1_epi32 (0), c; + __m128i a = _mm_set1_epi32 (param), b = _mm_set1_epi32 (param + 1), c; c = _mm_xor_si128 (a, b); - return 0; + return _mm_cvtsi128_si32(c); } _ACEOF if ac_fn_c_try_compile "$LINENO"; then : @@ -12870,10 +12930,11 @@ cat confdefs.h - <<_ACEOF >conftest.$ac_ext #include #include #include +int param; int main () { - __m128i a = _mm_set1_epi32 (0), b = _mm_set1_epi32 (0), c; + __m128i a = _mm_set1_epi32 (param), b = _mm_set1_epi32 (param + 1), c; c = _mm_maddubs_epi16 (a, b); - return 0; + return _mm_cvtsi128_si32(c); } _ACEOF if ac_fn_c_try_compile "$LINENO"; then : @@ -13920,6 +13981,15 @@ $as_echo "#define HAVE_FEENABLEEXCEPT 1" >>confdefs.h fi +ac_fn_c_check_decl "$LINENO" "FE_DIVBYZERO" "ac_cv_have_decl_FE_DIVBYZERO" "#include +" +if test "x$ac_cv_have_decl_FE_DIVBYZERO" = xyes; then : + +$as_echo "#define HAVE_FEDIVBYZERO 1" >>confdefs.h + +fi + + ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday" if test "x$ac_cv_func_gettimeofday" = xyes; then : have_gettimeofday=yes @@ -15259,7 +15329,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1 # report actual input values of CONFIG_FILES etc. instead of their # values after options handling. ac_log=" -This file was extended by pixman $as_me 0.32.8, which was +This file was extended by pixman $as_me 0.34.0, which was generated by GNU Autoconf 2.69. Invocation command line was CONFIG_FILES = $CONFIG_FILES @@ -15325,7 +15395,7 @@ _ACEOF cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1 ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`" ac_cs_version="\\ -pixman config.status 0.32.8 +pixman config.status 0.34.0 configured by $0, generated by GNU Autoconf 2.69, with options \\"\$ac_cs_config\\" diff --git a/lib/pixman/configure.ac b/lib/pixman/configure.ac index 40fcb8bb2..09991f61a 100644 --- a/lib/pixman/configure.ac +++ b/lib/pixman/configure.ac @@ -53,8 +53,8 @@ AC_PREREQ([2.57]) # m4_define([pixman_major], 0) -m4_define([pixman_minor], 32) -m4_define([pixman_micro], 8) +m4_define([pixman_minor], 34) +m4_define([pixman_micro], 0) m4_define([pixman_version],[pixman_major.pixman_minor.pixman_micro]) @@ -184,6 +184,7 @@ AC_SUBST(LT_VERSION_INFO) PIXMAN_CHECK_CFLAG([-Wall]) PIXMAN_CHECK_CFLAG([-Wdeclaration-after-statement]) +PIXMAN_CHECK_CFLAG([-Wno-unused-local-typedefs]) PIXMAN_CHECK_CFLAG([-fno-strict-aliasing]) dnl ========================================================================= @@ -346,15 +347,27 @@ AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ #error "Need GCC >= 3.4 for MMX intrinsics" #endif #include +#include + +/* Check support for block expressions */ +#define _mm_shuffle_pi16(A, N) \ + ({ \ + __m64 ret; \ + \ + /* Some versions of clang will choke on K */ \ + asm ("pshufw %2, %1, %0\n\t" \ + : "=y" (ret) \ + : "y" (A), "K" ((const int8_t)N) \ + ); \ + \ + ret; \ + }) + int main () { __m64 v = _mm_cvtsi32_si64 (1); __m64 w; - /* Some versions of clang will choke on K */ - asm ("pshufw %2, %1, %0\n\t" - : "=y" (w) - : "y" (v), "K" (5) - ); + w = _mm_shuffle_pi16(v, 5); /* Some versions of clang will choke on this */ asm ("pmulhuw %1, %0\n\t" @@ -416,10 +429,11 @@ AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ #include #include #include +int param; int main () { - __m128i a = _mm_set1_epi32 (0), b = _mm_set1_epi32 (0), c; + __m128i a = _mm_set1_epi32 (param), b = _mm_set1_epi32 (param + 1), c; c = _mm_xor_si128 (a, b); - return 0; + return _mm_cvtsi128_si32(c); }]])], have_sse2_intrinsics=yes) CFLAGS=$xserver_save_CFLAGS @@ -460,10 +474,11 @@ AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ #include #include #include +int param; int main () { - __m128i a = _mm_set1_epi32 (0), b = _mm_set1_epi32 (0), c; + __m128i a = _mm_set1_epi32 (param), b = _mm_set1_epi32 (param + 1), c; c = _mm_maddubs_epi16 (a, b); - return 0; + return _mm_cvtsi128_si32(c); }]])], have_ssse3_intrinsics=yes) CFLAGS=$xserver_save_CFLAGS @@ -890,6 +905,11 @@ if test x$have_feenableexcept = xyes; then AC_DEFINE(HAVE_FEENABLEEXCEPT, 1, [Whether we have feenableexcept()]) fi +AC_CHECK_DECL([FE_DIVBYZERO], + [AC_DEFINE(HAVE_FEDIVBYZERO, 1, [Whether we have FE_DIVBYZERO])], + [], + [[#include ]]) + AC_CHECK_FUNC(gettimeofday, have_gettimeofday=yes, have_gettimeofday=no) AC_CHECK_HEADER(sys/time.h, have_sys_time_h=yes, have_sys_time_h=no) if test x$have_gettimeofday = xyes && test x$have_sys_time_h = xyes; then diff --git a/lib/pixman/pixman/Makefile.am b/lib/pixman/pixman/Makefile.am index b376d9aeb..581b6f61e 100644 --- a/lib/pixman/pixman/Makefile.am +++ b/lib/pixman/pixman/Makefile.am @@ -72,6 +72,7 @@ libpixman_arm_simd_la_SOURCES = \ pixman-arm-common.h \ pixman-arm-simd-asm.S \ pixman-arm-simd-asm-scaled.S \ + pixman-arm-asm.h \ pixman-arm-simd-asm.h libpixman_1_la_LIBADD += libpixman-arm-simd.la @@ -86,6 +87,7 @@ libpixman_arm_neon_la_SOURCES = \ pixman-arm-common.h \ pixman-arm-neon-asm.S \ pixman-arm-neon-asm-bilinear.S \ + pixman-arm-asm.h \ pixman-arm-neon-asm.h libpixman_1_la_LIBADD += libpixman-arm-neon.la diff --git a/lib/pixman/pixman/Makefile.in b/lib/pixman/pixman/Makefile.in index 26c5fbe4f..db2fb9882 100644 --- a/lib/pixman/pixman/Makefile.in +++ b/lib/pixman/pixman/Makefile.in @@ -160,7 +160,8 @@ libpixman_1_la_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CC \ libpixman_arm_neon_la_LIBADD = am__libpixman_arm_neon_la_SOURCES_DIST = pixman-arm-neon.c \ pixman-arm-common.h pixman-arm-neon-asm.S \ - pixman-arm-neon-asm-bilinear.S pixman-arm-neon-asm.h + pixman-arm-neon-asm-bilinear.S pixman-arm-asm.h \ + pixman-arm-neon-asm.h @USE_ARM_NEON_TRUE@am_libpixman_arm_neon_la_OBJECTS = \ @USE_ARM_NEON_TRUE@ pixman-arm-neon.lo pixman-arm-neon-asm.lo \ @USE_ARM_NEON_TRUE@ pixman-arm-neon-asm-bilinear.lo @@ -169,7 +170,8 @@ libpixman_arm_neon_la_OBJECTS = $(am_libpixman_arm_neon_la_OBJECTS) libpixman_arm_simd_la_LIBADD = am__libpixman_arm_simd_la_SOURCES_DIST = pixman-arm-simd.c \ pixman-arm-common.h pixman-arm-simd-asm.S \ - pixman-arm-simd-asm-scaled.S pixman-arm-simd-asm.h + pixman-arm-simd-asm-scaled.S pixman-arm-asm.h \ + pixman-arm-simd-asm.h @USE_ARM_SIMD_TRUE@am_libpixman_arm_simd_la_OBJECTS = \ @USE_ARM_SIMD_TRUE@ pixman-arm-simd.lo pixman-arm-simd-asm.lo \ @USE_ARM_SIMD_TRUE@ pixman-arm-simd-asm-scaled.lo @@ -547,6 +549,7 @@ EXTRA_DIST = \ @USE_ARM_SIMD_TRUE@ pixman-arm-common.h \ @USE_ARM_SIMD_TRUE@ pixman-arm-simd-asm.S \ @USE_ARM_SIMD_TRUE@ pixman-arm-simd-asm-scaled.S \ +@USE_ARM_SIMD_TRUE@ pixman-arm-asm.h \ @USE_ARM_SIMD_TRUE@ pixman-arm-simd-asm.h @USE_ARM_SIMD_TRUE@ASM_CFLAGS_arm_simd = @@ -555,6 +558,7 @@ EXTRA_DIST = \ @USE_ARM_NEON_TRUE@ pixman-arm-common.h \ @USE_ARM_NEON_TRUE@ pixman-arm-neon-asm.S \ @USE_ARM_NEON_TRUE@ pixman-arm-neon-asm-bilinear.S \ +@USE_ARM_NEON_TRUE@ pixman-arm-asm.h \ @USE_ARM_NEON_TRUE@ pixman-arm-neon-asm.h @USE_ARM_NEON_TRUE@ASM_CFLAGS_arm_neon = diff --git a/lib/pixman/pixman/pixman-arm-asm.h b/lib/pixman/pixman/pixman-arm-asm.h new file mode 100644 index 000000000..ee7854108 --- /dev/null +++ b/lib/pixman/pixman/pixman-arm-asm.h @@ -0,0 +1,37 @@ +/* + * Copyright © 2008 Mozilla Corporation + * Copyright © 2010 Nokia Corporation + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Mozilla Corporation not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Mozilla Corporation makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Jeff Muizelaar (jeff@infidigm.net) + * + */ + +/* Supplementary macro for setting function attributes */ +.macro pixman_asm_function fname + .func fname + .global fname +#ifdef __ELF__ + .hidden fname + .type fname, %function +#endif +fname: +.endm diff --git a/lib/pixman/pixman/pixman-arm-common.h b/lib/pixman/pixman/pixman-arm-common.h index 3a7cb2bef..953768830 100644 --- a/lib/pixman/pixman/pixman-arm-common.h +++ b/lib/pixman/pixman/pixman-arm-common.h @@ -266,13 +266,6 @@ FAST_NEAREST_MAINLOOP (cputype##_##name##_normal_##op, \ scaled_nearest_scanline_##cputype##_##name##_##op, \ src_type, dst_type, NORMAL) -/* Provide entries for the fast path table */ -#define PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH(op,s,d,func) \ - SIMPLE_NEAREST_FAST_PATH_COVER (op,s,d,func), \ - SIMPLE_NEAREST_FAST_PATH_NONE (op,s,d,func), \ - SIMPLE_NEAREST_FAST_PATH_PAD (op,s,d,func), \ - SIMPLE_NEAREST_FAST_PATH_NORMAL (op,s,d,func) - #define PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_A8_DST(flags, cputype, name, op, \ src_type, dst_type) \ void \ @@ -318,9 +311,7 @@ FAST_NEAREST_MAINLOOP_COMMON (cputype##_##name##_normal_##op, \ /* Provide entries for the fast path table */ #define PIXMAN_ARM_SIMPLE_NEAREST_A8_MASK_FAST_PATH(op,s,d,func) \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_COVER (op,s,d,func), \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_NONE (op,s,d,func), \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_PAD (op,s,d,func), \ + SIMPLE_NEAREST_A8_MASK_FAST_PATH (op,s,d,func), \ SIMPLE_NEAREST_A8_MASK_FAST_PATH_NORMAL (op,s,d,func) /*****************************************************************************/ diff --git a/lib/pixman/pixman/pixman-arm-neon-asm-bilinear.S b/lib/pixman/pixman/pixman-arm-neon-asm-bilinear.S index e37b5c298..0fd92d61c 100644 --- a/lib/pixman/pixman/pixman-arm-neon-asm-bilinear.S +++ b/lib/pixman/pixman/pixman-arm-neon-asm-bilinear.S @@ -65,23 +65,13 @@ .p2align 2 #include "pixman-private.h" +#include "pixman-arm-asm.h" #include "pixman-arm-neon-asm.h" /* * Bilinear macros from pixman-arm-neon-asm.S */ -/* Supplementary macro for setting function attributes */ -.macro pixman_asm_function fname - .func fname - .global fname -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif -fname: -.endm - /* * Bilinear scaling support code which tries to provide pixel fetching, color * format conversion, and interpolation as separate macros which can be used diff --git a/lib/pixman/pixman/pixman-arm-neon-asm.S b/lib/pixman/pixman/pixman-arm-neon-asm.S index 187197dc3..7e949a38f 100644 --- a/lib/pixman/pixman/pixman-arm-neon-asm.S +++ b/lib/pixman/pixman/pixman-arm-neon-asm.S @@ -50,6 +50,7 @@ .p2align 2 #include "pixman-private.h" +#include "pixman-arm-asm.h" #include "pixman-arm-neon-asm.h" /* Global configuration options and preferences */ @@ -2830,17 +2831,6 @@ generate_composite_function_nearest_scanline \ /******************************************************************************/ -/* Supplementary macro for setting function attributes */ -.macro pixman_asm_function fname - .func fname - .global fname -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif -fname: -.endm - /* * Bilinear scaling support code which tries to provide pixel fetching, color * format conversion, and interpolation as separate macros which can be used diff --git a/lib/pixman/pixman/pixman-arm-neon-asm.h b/lib/pixman/pixman/pixman-arm-neon-asm.h index d0d92d74c..bdcf6a9d4 100644 --- a/lib/pixman/pixman/pixman-arm-neon-asm.h +++ b/lib/pixman/pixman/pixman-arm-neon-asm.h @@ -631,14 +631,8 @@ local skip1 src_basereg_ = 0, \ mask_basereg_ = 24 - .func fname - .global fname - /* For ELF format also set function visibility to hidden */ -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif -fname: + pixman_asm_function fname + push {r4-r12, lr} /* save all registers */ /* @@ -945,14 +939,8 @@ fname: src_basereg_ = 0, \ mask_basereg_ = 24 - .func fname - .global fname - /* For ELF format also set function visibility to hidden */ -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif -fname: + pixman_asm_function fname + .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_NONE /* * Make some macro arguments globally visible and accessible diff --git a/lib/pixman/pixman/pixman-arm-neon.c b/lib/pixman/pixman/pixman-arm-neon.c index 60e9c78d2..be761c965 100644 --- a/lib/pixman/pixman/pixman-arm-neon.c +++ b/lib/pixman/pixman/pixman-arm-neon.c @@ -362,21 +362,21 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8r8g8b8, neon_composite_out_reverse_8_8888), PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8b8g8r8, neon_composite_out_reverse_8_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, neon_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, neon_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, neon_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, neon_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, neon_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, neon_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, neon_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, neon_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, r5g6b5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, b5g6r5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, r5g6b5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, b5g6r5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, r5g6b5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, r5g6b5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, b5g6r5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, b5g6r5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, r5g6b5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, r5g6b5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, b5g6r5, neon_8888_0565), + SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, b5g6r5, neon_8888_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, b5g6r5, x8b8g8r8, neon_0565_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, x8r8g8b8, neon_0565_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, b5g6r5, x8b8g8r8, neon_0565_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, x8r8g8b8, neon_0565_8888), /* Note: NONE repeat is not supported yet */ SIMPLE_NEAREST_FAST_PATH_COVER (SRC, r5g6b5, a8r8g8b8, neon_0565_8888), SIMPLE_NEAREST_FAST_PATH_COVER (SRC, b5g6r5, a8b8g8r8, neon_0565_8888), diff --git a/lib/pixman/pixman/pixman-arm-simd-asm-scaled.S b/lib/pixman/pixman/pixman-arm-simd-asm-scaled.S index 711099548..e050292e0 100644 --- a/lib/pixman/pixman/pixman-arm-simd-asm-scaled.S +++ b/lib/pixman/pixman/pixman-arm-simd-asm-scaled.S @@ -37,16 +37,7 @@ .altmacro .p2align 2 -/* Supplementary macro for setting function attributes */ -.macro pixman_asm_function fname - .func fname - .global fname -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif -fname: -.endm +#include "pixman-arm-asm.h" /* * Note: This code is only using armv5te instructions (not even armv6), diff --git a/lib/pixman/pixman/pixman-arm-simd-asm.S b/lib/pixman/pixman/pixman-arm-simd-asm.S index c20968879..a74a0a8f3 100644 --- a/lib/pixman/pixman/pixman-arm-simd-asm.S +++ b/lib/pixman/pixman/pixman-arm-simd-asm.S @@ -37,6 +37,7 @@ .altmacro .p2align 2 +#include "pixman-arm-asm.h" #include "pixman-arm-simd-asm.h" /* A head macro should do all processing which results in an output of up to @@ -303,6 +304,83 @@ generate_composite_function \ /******************************************************************************/ +.macro src_x888_0565_init + /* Hold loop invariant in MASK */ + ldr MASK, =0x001F001F + line_saved_regs STRIDE_S, ORIG_W +.endm + +.macro src_x888_0565_1pixel s, d + and WK&d, MASK, WK&s, lsr #3 @ 00000000000rrrrr00000000000bbbbb + and STRIDE_S, WK&s, #0xFC00 @ 0000000000000000gggggg0000000000 + orr WK&d, WK&d, WK&d, lsr #5 @ 00000000000-----rrrrr000000bbbbb + orr WK&d, WK&d, STRIDE_S, lsr #5 @ 00000000000-----rrrrrggggggbbbbb + /* Top 16 bits are discarded during the following STRH */ +.endm + +.macro src_x888_0565_2pixels slo, shi, d, tmp + and SCRATCH, WK&shi, #0xFC00 @ 0000000000000000GGGGGG0000000000 + and WK&tmp, MASK, WK&shi, lsr #3 @ 00000000000RRRRR00000000000BBBBB + and WK&shi, MASK, WK&slo, lsr #3 @ 00000000000rrrrr00000000000bbbbb + orr WK&tmp, WK&tmp, WK&tmp, lsr #5 @ 00000000000-----RRRRR000000BBBBB + orr WK&tmp, WK&tmp, SCRATCH, lsr #5 @ 00000000000-----RRRRRGGGGGGBBBBB + and SCRATCH, WK&slo, #0xFC00 @ 0000000000000000gggggg0000000000 + orr WK&shi, WK&shi, WK&shi, lsr #5 @ 00000000000-----rrrrr000000bbbbb + orr WK&shi, WK&shi, SCRATCH, lsr #5 @ 00000000000-----rrrrrggggggbbbbb + pkhbt WK&d, WK&shi, WK&tmp, lsl #16 @ RRRRRGGGGGGBBBBBrrrrrggggggbbbbb +.endm + +.macro src_x888_0565_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + WK4 .req STRIDE_S + WK5 .req STRIDE_M + WK6 .req WK3 + WK7 .req ORIG_W + .if numbytes == 16 + pixld , 16, 4, SRC, 0 + src_x888_0565_2pixels 4, 5, 0, 0 + pixld , 8, 4, SRC, 0 + src_x888_0565_2pixels 6, 7, 1, 1 + pixld , 8, 6, SRC, 0 + .else + pixld , numbytes*2, 4, SRC, 0 + .endif +.endm + +.macro src_x888_0565_process_tail cond, numbytes, firstreg + .if numbytes == 16 + src_x888_0565_2pixels 4, 5, 2, 2 + src_x888_0565_2pixels 6, 7, 3, 4 + .elseif numbytes == 8 + src_x888_0565_2pixels 4, 5, 1, 1 + src_x888_0565_2pixels 6, 7, 2, 2 + .elseif numbytes == 4 + src_x888_0565_2pixels 4, 5, 1, 1 + .else + src_x888_0565_1pixel 4, 1 + .endif + .if numbytes == 16 + pixst , numbytes, 0, DST + .else + pixst , numbytes, 1, DST + .endif + .unreq WK4 + .unreq WK5 + .unreq WK6 + .unreq WK7 +.endm + +generate_composite_function \ + pixman_composite_src_x888_0565_asm_armv6, 32, 0, 16, \ + FLAG_DST_WRITEONLY | FLAG_BRANCH_OVER | FLAG_PROCESS_DOES_STORE | FLAG_SPILL_LINE_VARS | FLAG_PROCESS_CORRUPTS_SCRATCH, \ + 3, /* prefetch distance */ \ + src_x888_0565_init, \ + nop_macro, /* newline */ \ + nop_macro, /* cleanup */ \ + src_x888_0565_process_head, \ + src_x888_0565_process_tail + +/******************************************************************************/ + .macro add_8_8_8pixels cond, dst1, dst2 uqadd8&cond WK&dst1, WK&dst1, MASK uqadd8&cond WK&dst2, WK&dst2, STRIDE_M @@ -611,3 +689,491 @@ generate_composite_function \ /******************************************************************************/ +.macro over_reverse_n_8888_init + ldr SRC, [sp, #ARGS_STACK_OFFSET] + ldr MASK, =0x00800080 + /* Split source pixel into RB/AG parts */ + uxtb16 STRIDE_S, SRC + uxtb16 STRIDE_M, SRC, ror #8 + /* Set GE[3:0] to 0101 so SEL instructions do what we want */ + uadd8 SCRATCH, MASK, MASK + line_saved_regs STRIDE_D, ORIG_W +.endm + +.macro over_reverse_n_8888_newline + mov STRIDE_D, #0xFF +.endm + +.macro over_reverse_n_8888_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + pixld , numbytes, firstreg, DST, 0 +.endm + +.macro over_reverse_n_8888_1pixel d, is_only + teq WK&d, #0 + beq 8f /* replace with source */ + bics ORIG_W, STRIDE_D, WK&d, lsr #24 + .if is_only == 1 + beq 49f /* skip store */ + .else + beq 9f /* write same value back */ + .endif + mla SCRATCH, STRIDE_S, ORIG_W, MASK /* red/blue */ + mla ORIG_W, STRIDE_M, ORIG_W, MASK /* alpha/green */ + uxtab16 SCRATCH, SCRATCH, SCRATCH, ror #8 + uxtab16 ORIG_W, ORIG_W, ORIG_W, ror #8 + mov SCRATCH, SCRATCH, ror #8 + sel ORIG_W, SCRATCH, ORIG_W + uqadd8 WK&d, WK&d, ORIG_W + b 9f +8: mov WK&d, SRC +9: +.endm + +.macro over_reverse_n_8888_tail numbytes, reg1, reg2, reg3, reg4 + .if numbytes == 4 + over_reverse_n_8888_1pixel reg1, 1 + .else + and SCRATCH, WK®1, WK®2 + .if numbytes == 16 + and SCRATCH, SCRATCH, WK®3 + and SCRATCH, SCRATCH, WK®4 + .endif + mvns SCRATCH, SCRATCH, asr #24 + beq 49f /* skip store if all opaque */ + over_reverse_n_8888_1pixel reg1, 0 + over_reverse_n_8888_1pixel reg2, 0 + .if numbytes == 16 + over_reverse_n_8888_1pixel reg3, 0 + over_reverse_n_8888_1pixel reg4, 0 + .endif + .endif + pixst , numbytes, reg1, DST +49: +.endm + +.macro over_reverse_n_8888_process_tail cond, numbytes, firstreg + over_reverse_n_8888_tail numbytes, firstreg, %(firstreg+1), %(firstreg+2), %(firstreg+3) +.endm + +generate_composite_function \ + pixman_composite_over_reverse_n_8888_asm_armv6, 0, 0, 32 \ + FLAG_DST_READWRITE | FLAG_BRANCH_OVER | FLAG_PROCESS_CORRUPTS_PSR | FLAG_PROCESS_DOES_STORE | FLAG_SPILL_LINE_VARS | FLAG_PROCESS_CORRUPTS_SCRATCH, \ + 3, /* prefetch distance */ \ + over_reverse_n_8888_init, \ + over_reverse_n_8888_newline, \ + nop_macro, /* cleanup */ \ + over_reverse_n_8888_process_head, \ + over_reverse_n_8888_process_tail + +/******************************************************************************/ + +.macro over_white_8888_8888_ca_init + HALF .req SRC + TMP0 .req STRIDE_D + TMP1 .req STRIDE_S + TMP2 .req STRIDE_M + TMP3 .req ORIG_W + WK4 .req SCRATCH + line_saved_regs STRIDE_D, STRIDE_M, ORIG_W + ldr SCRATCH, =0x800080 + mov HALF, #0x80 + /* Set GE[3:0] to 0101 so SEL instructions do what we want */ + uadd8 SCRATCH, SCRATCH, SCRATCH + .set DST_PRELOAD_BIAS, 8 +.endm + +.macro over_white_8888_8888_ca_cleanup + .set DST_PRELOAD_BIAS, 0 + .unreq HALF + .unreq TMP0 + .unreq TMP1 + .unreq TMP2 + .unreq TMP3 + .unreq WK4 +.endm + +.macro over_white_8888_8888_ca_combine m, d + uxtb16 TMP1, TMP0 /* rb_notmask */ + uxtb16 TMP2, d /* rb_dest; 1 stall follows */ + smlatt TMP3, TMP2, TMP1, HALF /* red */ + smlabb TMP2, TMP2, TMP1, HALF /* blue */ + uxtb16 TMP0, TMP0, ror #8 /* ag_notmask */ + uxtb16 TMP1, d, ror #8 /* ag_dest; 1 stall follows */ + smlatt d, TMP1, TMP0, HALF /* alpha */ + smlabb TMP1, TMP1, TMP0, HALF /* green */ + pkhbt TMP0, TMP2, TMP3, lsl #16 /* rb; 1 stall follows */ + pkhbt TMP1, TMP1, d, lsl #16 /* ag */ + uxtab16 TMP0, TMP0, TMP0, ror #8 + uxtab16 TMP1, TMP1, TMP1, ror #8 + mov TMP0, TMP0, ror #8 + sel d, TMP0, TMP1 + uqadd8 d, d, m /* d is a late result */ +.endm + +.macro over_white_8888_8888_ca_1pixel_head + pixld , 4, 1, MASK, 0 + pixld , 4, 3, DST, 0 +.endm + +.macro over_white_8888_8888_ca_1pixel_tail + mvn TMP0, WK1 + teq WK1, WK1, asr #32 + bne 01f + bcc 03f + mov WK3, WK1 + b 02f +01: over_white_8888_8888_ca_combine WK1, WK3 +02: pixst , 4, 3, DST +03: +.endm + +.macro over_white_8888_8888_ca_2pixels_head + pixld , 8, 1, MASK, 0 +.endm + +.macro over_white_8888_8888_ca_2pixels_tail + pixld , 8, 3, DST + mvn TMP0, WK1 + teq WK1, WK1, asr #32 + bne 01f + movcs WK3, WK1 + bcs 02f + teq WK2, #0 + beq 05f + b 02f +01: over_white_8888_8888_ca_combine WK1, WK3 +02: mvn TMP0, WK2 + teq WK2, WK2, asr #32 + bne 03f + movcs WK4, WK2 + b 04f +03: over_white_8888_8888_ca_combine WK2, WK4 +04: pixst , 8, 3, DST +05: +.endm + +.macro over_white_8888_8888_ca_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + .if numbytes == 4 + over_white_8888_8888_ca_1pixel_head + .else + .if numbytes == 16 + over_white_8888_8888_ca_2pixels_head + over_white_8888_8888_ca_2pixels_tail + .endif + over_white_8888_8888_ca_2pixels_head + .endif +.endm + +.macro over_white_8888_8888_ca_process_tail cond, numbytes, firstreg + .if numbytes == 4 + over_white_8888_8888_ca_1pixel_tail + .else + over_white_8888_8888_ca_2pixels_tail + .endif +.endm + +generate_composite_function \ + pixman_composite_over_white_8888_8888_ca_asm_armv6, 0, 32, 32 \ + FLAG_DST_READWRITE | FLAG_BRANCH_OVER | FLAG_PROCESS_CORRUPTS_PSR | FLAG_PROCESS_DOES_STORE | FLAG_SPILL_LINE_VARS | FLAG_PROCESS_CORRUPTS_SCRATCH \ + 2, /* prefetch distance */ \ + over_white_8888_8888_ca_init, \ + nop_macro, /* newline */ \ + over_white_8888_8888_ca_cleanup, \ + over_white_8888_8888_ca_process_head, \ + over_white_8888_8888_ca_process_tail + + +.macro over_n_8888_8888_ca_init + /* Set up constants. RB_SRC and AG_SRC are in registers; + * RB_FLDS, A_SRC, and the two HALF values need to go on the + * stack (and the ful SRC value is already there) */ + ldr SCRATCH, [sp, #ARGS_STACK_OFFSET] + mov WK0, #0x00FF0000 + orr WK0, WK0, #0xFF /* RB_FLDS (0x00FF00FF) */ + mov WK1, #0x80 /* HALF default value */ + mov WK2, SCRATCH, lsr #24 /* A_SRC */ + orr WK3, WK1, WK1, lsl #16 /* HALF alternate value (0x00800080) */ + push {WK0-WK3} + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET+16 + uxtb16 SRC, SCRATCH + uxtb16 STRIDE_S, SCRATCH, ror #8 + + /* Set GE[3:0] to 0101 so SEL instructions do what we want */ + uadd8 SCRATCH, WK3, WK3 + + .unreq WK0 + .unreq WK1 + .unreq WK2 + .unreq WK3 + WK0 .req Y + WK1 .req STRIDE_D + RB_SRC .req SRC + AG_SRC .req STRIDE_S + WK2 .req STRIDE_M + RB_FLDS .req r8 /* the reloaded constants have to be at consecutive registers starting at an even one */ + A_SRC .req r8 + HALF .req r9 + WK3 .req r10 + WK4 .req r11 + WK5 .req SCRATCH + WK6 .req ORIG_W + + line_saved_regs Y, STRIDE_D, STRIDE_M, ORIG_W +.endm + +.macro over_n_8888_8888_ca_cleanup + add sp, sp, #16 + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET-16 + + .unreq WK0 + .unreq WK1 + .unreq RB_SRC + .unreq AG_SRC + .unreq WK2 + .unreq RB_FLDS + .unreq A_SRC + .unreq HALF + .unreq WK3 + .unreq WK4 + .unreq WK5 + .unreq WK6 + WK0 .req r8 + WK1 .req r9 + WK2 .req r10 + WK3 .req r11 +.endm + +.macro over_n_8888_8888_ca_1pixel_head + pixld , 4, 6, MASK, 0 + pixld , 4, 0, DST, 0 +.endm + +.macro over_n_8888_8888_ca_1pixel_tail + ldrd A_SRC, HALF, [sp, #LOCALS_STACK_OFFSET+8] + uxtb16 WK1, WK6 /* rb_mask (first step of hard case placed in what would otherwise be a stall) */ + teq WK6, WK6, asr #32 /* Zc if transparent, ZC if opaque */ + bne 20f + bcc 40f + /* Mask is fully opaque (all channels) */ + ldr WK6, [sp, #ARGS_STACK_OFFSET] /* get SRC back */ + eors A_SRC, A_SRC, #0xFF + bne 10f + /* Source is also opaque - same as src_8888_8888 */ + mov WK0, WK6 + b 30f +10: /* Same as over_8888_8888 */ + mul_8888_8 WK0, A_SRC, WK5, HALF + uqadd8 WK0, WK0, WK6 + b 30f +20: /* No simplifications possible - do it the hard way */ + uxtb16 WK2, WK6, ror #8 /* ag_mask */ + mla WK3, WK1, A_SRC, HALF /* rb_mul; 2 cycles */ + mla WK4, WK2, A_SRC, HALF /* ag_mul; 2 cycles */ + ldrd RB_FLDS, HALF, [sp, #LOCALS_STACK_OFFSET] + uxtb16 WK5, WK0 /* rb_dest */ + uxtab16 WK3, WK3, WK3, ror #8 + uxtb16 WK6, WK0, ror #8 /* ag_dest */ + uxtab16 WK4, WK4, WK4, ror #8 + smlatt WK0, RB_SRC, WK1, HALF /* red1 */ + smlabb WK1, RB_SRC, WK1, HALF /* blue1 */ + bic WK3, RB_FLDS, WK3, lsr #8 + bic WK4, RB_FLDS, WK4, lsr #8 + pkhbt WK1, WK1, WK0, lsl #16 /* rb1 */ + smlatt WK0, WK5, WK3, HALF /* red2 */ + smlabb WK3, WK5, WK3, HALF /* blue2 */ + uxtab16 WK1, WK1, WK1, ror #8 + smlatt WK5, AG_SRC, WK2, HALF /* alpha1 */ + pkhbt WK3, WK3, WK0, lsl #16 /* rb2 */ + smlabb WK0, AG_SRC, WK2, HALF /* green1 */ + smlatt WK2, WK6, WK4, HALF /* alpha2 */ + smlabb WK4, WK6, WK4, HALF /* green2 */ + pkhbt WK0, WK0, WK5, lsl #16 /* ag1 */ + uxtab16 WK3, WK3, WK3, ror #8 + pkhbt WK4, WK4, WK2, lsl #16 /* ag2 */ + uxtab16 WK0, WK0, WK0, ror #8 + uxtab16 WK4, WK4, WK4, ror #8 + mov WK1, WK1, ror #8 + mov WK3, WK3, ror #8 + sel WK2, WK1, WK0 /* recombine source*mask */ + sel WK1, WK3, WK4 /* recombine dest*(1-source_alpha*mask) */ + uqadd8 WK0, WK1, WK2 /* followed by 1 stall */ +30: /* The destination buffer is already in the L1 cache, so + * there's little point in amalgamating writes */ + pixst , 4, 0, DST +40: +.endm + +.macro over_n_8888_8888_ca_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + .rept (numbytes / 4) - 1 + over_n_8888_8888_ca_1pixel_head + over_n_8888_8888_ca_1pixel_tail + .endr + over_n_8888_8888_ca_1pixel_head +.endm + +.macro over_n_8888_8888_ca_process_tail cond, numbytes, firstreg + over_n_8888_8888_ca_1pixel_tail +.endm + +pixman_asm_function pixman_composite_over_n_8888_8888_ca_asm_armv6 + ldr ip, [sp] + cmp ip, #-1 + beq pixman_composite_over_white_8888_8888_ca_asm_armv6 + /* else drop through... */ + .endfunc +generate_composite_function \ + pixman_composite_over_n_8888_8888_ca_asm_armv6_helper, 0, 32, 32 \ + FLAG_DST_READWRITE | FLAG_BRANCH_OVER | FLAG_PROCESS_CORRUPTS_PSR | FLAG_PROCESS_DOES_STORE | FLAG_SPILL_LINE_VARS | FLAG_PROCESS_CORRUPTS_SCRATCH | FLAG_PROCESS_CORRUPTS_WK0 \ + 2, /* prefetch distance */ \ + over_n_8888_8888_ca_init, \ + nop_macro, /* newline */ \ + over_n_8888_8888_ca_cleanup, \ + over_n_8888_8888_ca_process_head, \ + over_n_8888_8888_ca_process_tail + +/******************************************************************************/ + +.macro in_reverse_8888_8888_init + /* Hold loop invariant in MASK */ + ldr MASK, =0x00800080 + /* Set GE[3:0] to 0101 so SEL instructions do what we want */ + uadd8 SCRATCH, MASK, MASK + /* Offset the source pointer: we only need the alpha bytes */ + add SRC, SRC, #3 + line_saved_regs ORIG_W +.endm + +.macro in_reverse_8888_8888_head numbytes, reg1, reg2, reg3 + ldrb ORIG_W, [SRC], #4 + .if numbytes >= 8 + ldrb WK®1, [SRC], #4 + .if numbytes == 16 + ldrb WK®2, [SRC], #4 + ldrb WK®3, [SRC], #4 + .endif + .endif + add DST, DST, #numbytes +.endm + +.macro in_reverse_8888_8888_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + in_reverse_8888_8888_head numbytes, firstreg, %(firstreg+1), %(firstreg+2) +.endm + +.macro in_reverse_8888_8888_1pixel s, d, offset, is_only + .if is_only != 1 + movs s, ORIG_W + .if offset != 0 + ldrb ORIG_W, [SRC, #offset] + .endif + beq 01f + teq STRIDE_M, #0xFF + beq 02f + .endif + uxtb16 SCRATCH, d /* rb_dest */ + uxtb16 d, d, ror #8 /* ag_dest */ + mla SCRATCH, SCRATCH, s, MASK + mla d, d, s, MASK + uxtab16 SCRATCH, SCRATCH, SCRATCH, ror #8 + uxtab16 d, d, d, ror #8 + mov SCRATCH, SCRATCH, ror #8 + sel d, SCRATCH, d + b 02f + .if offset == 0 +48: /* Last mov d,#0 of the set - used as part of shortcut for + * source values all 0 */ + .endif +01: mov d, #0 +02: +.endm + +.macro in_reverse_8888_8888_tail numbytes, reg1, reg2, reg3, reg4 + .if numbytes == 4 + teq ORIG_W, ORIG_W, asr #32 + ldrne WK®1, [DST, #-4] + .elseif numbytes == 8 + teq ORIG_W, WK®1 + teqeq ORIG_W, ORIG_W, asr #32 /* all 0 or all -1? */ + ldmnedb DST, {WK®1-WK®2} + .else + teq ORIG_W, WK®1 + teqeq ORIG_W, WK®2 + teqeq ORIG_W, WK®3 + teqeq ORIG_W, ORIG_W, asr #32 /* all 0 or all -1? */ + ldmnedb DST, {WK®1-WK®4} + .endif + cmnne DST, #0 /* clear C if NE */ + bcs 49f /* no writes to dest if source all -1 */ + beq 48f /* set dest to all 0 if source all 0 */ + .if numbytes == 4 + in_reverse_8888_8888_1pixel ORIG_W, WK®1, 0, 1 + str WK®1, [DST, #-4] + .elseif numbytes == 8 + in_reverse_8888_8888_1pixel STRIDE_M, WK®1, -4, 0 + in_reverse_8888_8888_1pixel STRIDE_M, WK®2, 0, 0 + stmdb DST, {WK®1-WK®2} + .else + in_reverse_8888_8888_1pixel STRIDE_M, WK®1, -12, 0 + in_reverse_8888_8888_1pixel STRIDE_M, WK®2, -8, 0 + in_reverse_8888_8888_1pixel STRIDE_M, WK®3, -4, 0 + in_reverse_8888_8888_1pixel STRIDE_M, WK®4, 0, 0 + stmdb DST, {WK®1-WK®4} + .endif +49: +.endm + +.macro in_reverse_8888_8888_process_tail cond, numbytes, firstreg + in_reverse_8888_8888_tail numbytes, firstreg, %(firstreg+1), %(firstreg+2), %(firstreg+3) +.endm + +generate_composite_function \ + pixman_composite_in_reverse_8888_8888_asm_armv6, 32, 0, 32 \ + FLAG_DST_READWRITE | FLAG_BRANCH_OVER | FLAG_PROCESS_CORRUPTS_PSR | FLAG_PROCESS_DOES_STORE | FLAG_SPILL_LINE_VARS | FLAG_PROCESS_CORRUPTS_SCRATCH | FLAG_NO_PRELOAD_DST \ + 2, /* prefetch distance */ \ + in_reverse_8888_8888_init, \ + nop_macro, /* newline */ \ + nop_macro, /* cleanup */ \ + in_reverse_8888_8888_process_head, \ + in_reverse_8888_8888_process_tail + +/******************************************************************************/ + +.macro over_n_8888_init + ldr SRC, [sp, #ARGS_STACK_OFFSET] + /* Hold loop invariant in MASK */ + ldr MASK, =0x00800080 + /* Hold multiplier for destination in STRIDE_M */ + mov STRIDE_M, #255 + sub STRIDE_M, STRIDE_M, SRC, lsr #24 + /* Set GE[3:0] to 0101 so SEL instructions do what we want */ + uadd8 SCRATCH, MASK, MASK +.endm + +.macro over_n_8888_process_head cond, numbytes, firstreg, unaligned_src, unaligned_mask, preload + pixld , numbytes, firstreg, DST, 0 +.endm + +.macro over_n_8888_1pixel dst + mul_8888_8 WK&dst, STRIDE_M, SCRATCH, MASK + uqadd8 WK&dst, WK&dst, SRC +.endm + +.macro over_n_8888_process_tail cond, numbytes, firstreg + .set PROCESS_REG, firstreg + .rept numbytes / 4 + over_n_8888_1pixel %(PROCESS_REG) + .set PROCESS_REG, PROCESS_REG+1 + .endr + pixst , numbytes, firstreg, DST +.endm + +generate_composite_function \ + pixman_composite_over_n_8888_asm_armv6, 0, 0, 32 \ + FLAG_DST_READWRITE | FLAG_BRANCH_OVER | FLAG_PROCESS_DOES_STORE \ + 2, /* prefetch distance */ \ + over_n_8888_init, \ + nop_macro, /* newline */ \ + nop_macro, /* cleanup */ \ + over_n_8888_process_head, \ + over_n_8888_process_tail + +/******************************************************************************/ diff --git a/lib/pixman/pixman/pixman-arm-simd-asm.h b/lib/pixman/pixman/pixman-arm-simd-asm.h index 65436062b..da153c3f5 100644 --- a/lib/pixman/pixman/pixman-arm-simd-asm.h +++ b/lib/pixman/pixman/pixman-arm-simd-asm.h @@ -76,6 +76,16 @@ .set FLAG_SPILL_LINE_VARS, 48 .set FLAG_PROCESS_CORRUPTS_SCRATCH, 0 .set FLAG_PROCESS_PRESERVES_SCRATCH, 64 +.set FLAG_PROCESS_PRESERVES_WK0, 0 +.set FLAG_PROCESS_CORRUPTS_WK0, 128 /* if possible, use the specified register(s) instead so WK0 can hold number of leading pixels */ +.set FLAG_PRELOAD_DST, 0 +.set FLAG_NO_PRELOAD_DST, 256 + +/* + * Number of bytes by which to adjust preload offset of destination + * buffer (allows preload instruction to be moved before the load(s)) + */ +.set DST_PRELOAD_BIAS, 0 /* * Offset into stack where mask and source pointer/stride can be accessed. @@ -86,6 +96,11 @@ .set ARGS_STACK_OFFSET, (9*4) #endif +/* + * Offset into stack where space allocated during init macro can be accessed. + */ +.set LOCALS_STACK_OFFSET, 0 + /* * Constants for selecting preferable prefetch type. */ @@ -196,8 +211,8 @@ PF add, SCRATCH, base, WK0, lsl #bpp_shift-dst_bpp_shift PF and, SCRATCH, SCRATCH, #31 PF rsb, SCRATCH, SCRATCH, WK0, lsl #bpp_shift-dst_bpp_shift - PF sub, SCRATCH, SCRATCH, #1 /* so now ranges are -16..-1 / 0..31 / 32..63 */ - PF movs, SCRATCH, SCRATCH, #32-6 /* so this sets NC / nc / Nc */ + PF sub, SCRATCH, SCRATCH, #1 /* so now ranges are -16..-1 / 0..31 / 32..63 */ + PF movs, SCRATCH, SCRATCH, lsl #32-6 /* so this sets NC / nc / Nc */ PF bcs, 61f PF bpl, 60f PF pld, [ptr, #32*(prefetch_distance+2)] @@ -359,23 +374,41 @@ .macro test_bits_1_0_ptr + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + movs SCRATCH, X, lsl #32-1 /* C,N = bits 1,0 of DST */ + .else movs SCRATCH, WK0, lsl #32-1 /* C,N = bits 1,0 of DST */ + .endif .endm .macro test_bits_3_2_ptr + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + movs SCRATCH, X, lsl #32-3 /* C,N = bits 3, 2 of DST */ + .else movs SCRATCH, WK0, lsl #32-3 /* C,N = bits 3, 2 of DST */ + .endif .endm .macro leading_15bytes process_head, process_tail /* On entry, WK0 bits 0-3 = number of bytes until destination is 16-byte aligned */ + .set DECREMENT_X, 1 + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + .set DECREMENT_X, 0 + sub X, X, WK0, lsr #dst_bpp_shift + str X, [sp, #LINE_SAVED_REG_COUNT*4] + mov X, WK0 + .endif /* Use unaligned loads in all cases for simplicity */ .if dst_w_bpp == 8 - conditional_process2 test_bits_1_0_ptr, mi, cs, process_head, process_tail, 1, 2, 1, 2, 1, 1, 1 + conditional_process2 test_bits_1_0_ptr, mi, cs, process_head, process_tail, 1, 2, 1, 2, 1, 1, DECREMENT_X .elseif dst_w_bpp == 16 test_bits_1_0_ptr - conditional_process1 cs, process_head, process_tail, 2, 2, 1, 1, 1 + conditional_process1 cs, process_head, process_tail, 2, 2, 1, 1, DECREMENT_X + .endif + conditional_process2 test_bits_3_2_ptr, mi, cs, process_head, process_tail, 4, 8, 1, 2, 1, 1, DECREMENT_X + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + ldr X, [sp, #LINE_SAVED_REG_COUNT*4] .endif - conditional_process2 test_bits_3_2_ptr, mi, cs, process_head, process_tail, 4, 8, 1, 2, 1, 1, 1 .endm .macro test_bits_3_2_pix @@ -414,7 +447,7 @@ preload_middle src_bpp, SRC, 0 preload_middle mask_bpp, MASK, 0 .endif - .if (dst_r_bpp > 0) && ((SUBBLOCK % 2) == 0) + .if (dst_r_bpp > 0) && ((SUBBLOCK % 2) == 0) && (((flags) & FLAG_NO_PRELOAD_DST) == 0) /* Because we know that writes are 16-byte aligned, it's relatively easy to ensure that * destination prefetches are 32-byte aligned. It's also the easiest channel to offset * preloads for, to achieve staggered prefetches for multiple channels, because there are @@ -437,11 +470,11 @@ .if dst_r_bpp > 0 tst DST, #16 bne 111f - process_inner_loop process_head, process_tail, unaligned_src, unaligned_mask, 16 + process_inner_loop process_head, process_tail, unaligned_src, unaligned_mask, 16 + DST_PRELOAD_BIAS b 112f 111: .endif - process_inner_loop process_head, process_tail, unaligned_src, unaligned_mask, 0 + process_inner_loop process_head, process_tail, unaligned_src, unaligned_mask, 0 + DST_PRELOAD_BIAS 112: /* Just before the final (prefetch_distance+1) 32-byte blocks, deal with final preloads */ .if (src_bpp*pix_per_block > 256) || (mask_bpp*pix_per_block > 256) || (dst_r_bpp*pix_per_block > 256) @@ -449,7 +482,9 @@ .endif preload_trailing src_bpp, src_bpp_shift, SRC preload_trailing mask_bpp, mask_bpp_shift, MASK + .if ((flags) & FLAG_NO_PRELOAD_DST) == 0 preload_trailing dst_r_bpp, dst_bpp_shift, DST + .endif add X, X, #(prefetch_distance+2)*pix_per_block - 128/dst_w_bpp /* The remainder of the line is handled identically to the medium case */ medium_case_inner_loop_and_trailing_pixels process_head, process_tail,, exit_label, unaligned_src, unaligned_mask @@ -561,13 +596,7 @@ process_tail, \ process_inner_loop - .func fname - .global fname - /* For ELF format also set function visibility to hidden */ -#ifdef __ELF__ - .hidden fname - .type fname, %function -#endif + pixman_asm_function fname /* * Make some macro arguments globally visible and accessible @@ -679,7 +708,6 @@ SCRATCH .req r12 ORIG_W .req r14 /* width (pixels) */ -fname: push {r4-r11, lr} /* save all registers */ subs Y, Y, #1 @@ -705,6 +733,13 @@ fname: #endif init + + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + /* Reserve a word in which to store X during leading pixels */ + sub sp, sp, #4 + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET+4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET+4 + .endif lsl STRIDE_D, #dst_bpp_shift /* stride in bytes */ sub STRIDE_D, STRIDE_D, X, lsl #dst_bpp_shift @@ -734,42 +769,49 @@ fname: .if (flags) & FLAG_SPILL_LINE_VARS_WIDE /* This is stmdb sp!,{} */ .word 0xE92D0000 | LINE_SAVED_REGS + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET + LINE_SAVED_REG_COUNT*4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET + LINE_SAVED_REG_COUNT*4 .endif 151: /* New line */ newline preload_leading_step1 src_bpp, WK1, SRC preload_leading_step1 mask_bpp, WK2, MASK + .if ((flags) & FLAG_NO_PRELOAD_DST) == 0 preload_leading_step1 dst_r_bpp, WK3, DST + .endif - tst DST, #15 + ands WK0, DST, #15 beq 154f - rsb WK0, DST, #0 /* bits 0-3 = number of leading bytes until destination aligned */ - .if (src_bpp != 0 && src_bpp != 2*dst_w_bpp) || (mask_bpp != 0 && mask_bpp != 2*dst_w_bpp) - PF and, WK0, WK0, #15 - .endif + rsb WK0, WK0, #16 /* number of leading bytes until destination aligned */ preload_leading_step2 src_bpp, src_bpp_shift, WK1, SRC preload_leading_step2 mask_bpp, mask_bpp_shift, WK2, MASK + .if ((flags) & FLAG_NO_PRELOAD_DST) == 0 preload_leading_step2 dst_r_bpp, dst_bpp_shift, WK3, DST + .endif leading_15bytes process_head, process_tail 154: /* Destination now 16-byte aligned; we have at least one prefetch on each channel as well as at least one 16-byte output block */ - .if (src_bpp > 0) && (mask_bpp == 0) && ((flags) & FLAG_PROCESS_PRESERVES_SCRATCH) + .if (src_bpp > 0) && (mask_bpp == 0) && ((flags) & FLAG_PROCESS_PRESERVES_SCRATCH) and SCRATCH, SRC, #31 rsb SCRATCH, SCRATCH, #32*prefetch_distance - .elseif (src_bpp == 0) && (mask_bpp > 0) && ((flags) & FLAG_PROCESS_PRESERVES_SCRATCH) + .elseif (src_bpp == 0) && (mask_bpp > 0) && ((flags) & FLAG_PROCESS_PRESERVES_SCRATCH) and SCRATCH, MASK, #31 rsb SCRATCH, SCRATCH, #32*prefetch_distance - .endif - .ifc "process_inner_loop","" + .endif + .ifc "process_inner_loop","" switch_on_alignment wide_case_inner_loop_and_trailing_pixels, process_head, process_tail, wide_case_inner_loop, 157f - .else + .else switch_on_alignment wide_case_inner_loop_and_trailing_pixels, process_head, process_tail, process_inner_loop, 157f - .endif + .endif 157: /* Check for another line */ end_of_line 1, %((flags) & FLAG_SPILL_LINE_VARS_WIDE), 151b + .if (flags) & FLAG_SPILL_LINE_VARS_WIDE + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET - LINE_SAVED_REG_COUNT*4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET - LINE_SAVED_REG_COUNT*4 + .endif .endif .ltorg @@ -779,17 +821,21 @@ fname: .if (flags) & FLAG_SPILL_LINE_VARS_NON_WIDE /* This is stmdb sp!,{} */ .word 0xE92D0000 | LINE_SAVED_REGS + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET + LINE_SAVED_REG_COUNT*4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET + LINE_SAVED_REG_COUNT*4 .endif 161: /* New line */ newline preload_line 0, src_bpp, src_bpp_shift, SRC /* in: X, corrupts: WK0-WK1 */ preload_line 0, mask_bpp, mask_bpp_shift, MASK + .if ((flags) & FLAG_NO_PRELOAD_DST) == 0 preload_line 0, dst_r_bpp, dst_bpp_shift, DST + .endif sub X, X, #128/dst_w_bpp /* simplifies inner loop termination */ - tst DST, #15 + ands WK0, DST, #15 beq 164f - rsb WK0, DST, #0 /* bits 0-3 = number of leading bytes until destination aligned */ + rsb WK0, WK0, #16 /* number of leading bytes until destination aligned */ leading_15bytes process_head, process_tail @@ -813,7 +859,9 @@ fname: newline preload_line 1, src_bpp, src_bpp_shift, SRC /* in: X, corrupts: WK0-WK1 */ preload_line 1, mask_bpp, mask_bpp_shift, MASK + .if ((flags) & FLAG_NO_PRELOAD_DST) == 0 preload_line 1, dst_r_bpp, dst_bpp_shift, DST + .endif .if dst_w_bpp == 8 tst DST, #3 @@ -844,12 +892,22 @@ fname: 177: /* Check for another line */ end_of_line %(dst_w_bpp < 32), %((flags) & FLAG_SPILL_LINE_VARS_NON_WIDE), 171b, last_one + .if (flags) & FLAG_SPILL_LINE_VARS_NON_WIDE + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET - LINE_SAVED_REG_COUNT*4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET - LINE_SAVED_REG_COUNT*4 + .endif 197: .if (flags) & FLAG_SPILL_LINE_VARS add sp, sp, #LINE_SAVED_REG_COUNT*4 .endif 198: + .if (flags) & FLAG_PROCESS_CORRUPTS_WK0 + .set ARGS_STACK_OFFSET, ARGS_STACK_OFFSET-4 + .set LOCALS_STACK_OFFSET, LOCALS_STACK_OFFSET-4 + add sp, sp, #4 + .endif + cleanup #ifdef DEBUG_PARAMS diff --git a/lib/pixman/pixman/pixman-arm-simd.c b/lib/pixman/pixman/pixman-arm-simd.c index af062e19d..f0d14540b 100644 --- a/lib/pixman/pixman/pixman-arm-simd.c +++ b/lib/pixman/pixman/pixman-arm-simd.c @@ -41,11 +41,20 @@ PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, src_8_8, uint8_t, 1, uint8_t, 1) PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, src_0565_8888, uint16_t, 1, uint32_t, 1) +PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, src_x888_0565, + uint32_t, 1, uint16_t, 1) PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, add_8_8, uint8_t, 1, uint8_t, 1) PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, over_8888_8888, uint32_t, 1, uint32_t, 1) +PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, in_reverse_8888_8888, + uint32_t, 1, uint32_t, 1) + +PIXMAN_ARM_BIND_FAST_PATH_N_DST (SKIP_ZERO_SRC, armv6, over_n_8888, + uint32_t, 1) +PIXMAN_ARM_BIND_FAST_PATH_N_DST (0, armv6, over_reverse_n_8888, + uint32_t, 1) PIXMAN_ARM_BIND_FAST_PATH_SRC_N_DST (SKIP_ZERO_MASK, armv6, over_8888_n_8888, uint32_t, 1, uint32_t, 1) @@ -53,6 +62,9 @@ PIXMAN_ARM_BIND_FAST_PATH_SRC_N_DST (SKIP_ZERO_MASK, armv6, over_8888_n_8888, PIXMAN_ARM_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, armv6, over_n_8_8888, uint8_t, 1, uint32_t, 1) +PIXMAN_ARM_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, armv6, over_n_8888_8888_ca, + uint32_t, 1, uint32_t, 1) + PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_DST (armv6, 0565_0565, SRC, uint16_t, uint16_t) PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_DST (armv6, 8888_8888, SRC, @@ -216,6 +228,11 @@ static const pixman_fast_path_t arm_simd_fast_paths[] = PIXMAN_STD_FAST_PATH (SRC, b5g6r5, null, a8b8g8r8, armv6_composite_src_0565_8888), PIXMAN_STD_FAST_PATH (SRC, b5g6r5, null, x8b8g8r8, armv6_composite_src_0565_8888), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, r5g6b5, armv6_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, r5g6b5, armv6_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, b5g6r5, armv6_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, b5g6r5, armv6_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, a8r8g8b8, armv6_composite_over_8888_8888), PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, armv6_composite_over_8888_8888), PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, armv6_composite_over_8888_8888), @@ -225,6 +242,13 @@ static const pixman_fast_path_t arm_simd_fast_paths[] = PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, a8b8g8r8, armv6_composite_over_8888_n_8888), PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, x8b8g8r8, armv6_composite_over_8888_n_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, null, a8r8g8b8, armv6_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, null, x8r8g8b8, armv6_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, null, a8b8g8r8, armv6_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, null, x8b8g8r8, armv6_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8r8g8b8, armv6_composite_over_reverse_n_8888), + PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8b8g8r8, armv6_composite_over_reverse_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, armv6_composite_add_8_8), PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, armv6_composite_over_n_8_8888), @@ -232,15 +256,25 @@ static const pixman_fast_path_t arm_simd_fast_paths[] = PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, armv6_composite_over_n_8_8888), PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, armv6_composite_over_n_8_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, r5g6b5, armv6_0565_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, b5g6r5, b5g6r5, armv6_0565_0565), + PIXMAN_STD_FAST_PATH (IN_REVERSE, a8r8g8b8, null, a8r8g8b8, armv6_composite_in_reverse_8888_8888), + PIXMAN_STD_FAST_PATH (IN_REVERSE, a8r8g8b8, null, x8r8g8b8, armv6_composite_in_reverse_8888_8888), + PIXMAN_STD_FAST_PATH (IN_REVERSE, a8b8g8r8, null, a8b8g8r8, armv6_composite_in_reverse_8888_8888), + PIXMAN_STD_FAST_PATH (IN_REVERSE, a8b8g8r8, null, x8b8g8r8, armv6_composite_in_reverse_8888_8888), + + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, armv6_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, armv6_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, armv6_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, x8b8g8r8, armv6_composite_over_n_8888_8888_ca), + + SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, r5g6b5, armv6_0565_0565), + SIMPLE_NEAREST_FAST_PATH (SRC, b5g6r5, b5g6r5, armv6_0565_0565), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, armv6_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, armv6_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, armv6_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, a8b8g8r8, armv6_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, x8b8g8r8, armv6_8888_8888), - PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, x8b8g8r8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, a8b8g8r8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, x8b8g8r8, armv6_8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, x8b8g8r8, armv6_8888_8888), { PIXMAN_OP_NONE }, }; diff --git a/lib/pixman/pixman/pixman-combine-float.c b/lib/pixman/pixman/pixman-combine-float.c index 5ea739f76..f5145bc9d 100644 --- a/lib/pixman/pixman/pixman-combine-float.c +++ b/lib/pixman/pixman/pixman-combine-float.c @@ -319,23 +319,44 @@ MAKE_PD_COMBINERS (conjoint_xor, ONE_MINUS_DA_OVER_SA, ONE_MINUS_SA_OVER_DA) * * The following blend modes have been taken from the PDF ISO 32000 * specification, which at this point in time is available from - * http://www.adobe.com/devnet/acrobat/pdfs/PDF32000_2008.pdf - * The relevant chapters are 11.3.5 and 11.3.6. + * + * http://www.adobe.com/devnet/pdf/pdf_reference.html + * + * The specific documents of interest are the PDF spec itself: + * + * http://wwwimages.adobe.com/www.adobe.com/content/dam/Adobe/en/devnet/pdf/pdfs/PDF32000_2008.pdf + * + * chapters 11.3.5 and 11.3.6 and a later supplement for Adobe Acrobat + * 9.1 and Reader 9.1: + * + * http://wwwimages.adobe.com/www.adobe.com/content/dam/Adobe/en/devnet/pdf/pdfs/adobe_supplement_iso32000_1.pdf + * + * that clarifies the specifications for blend modes ColorDodge and + * ColorBurn. + * * The formula for computing the final pixel color given in 11.3.6 is: - * αr × Cr = (1 – αs) × αb × Cb + (1 – αb) × αs × Cs + αb × αs × B(Cb, Cs) - * with B() being the blend function. - * Note that OVER is a special case of this operation, using B(Cb, Cs) = Cs - * - * These blend modes should match the SVG filter draft specification, as - * it has been designed to mirror ISO 32000. Note that at the current point - * no released draft exists that shows this, as the formulas have not been - * updated yet after the release of ISO 32000. - * - * The default implementation here uses the PDF_SEPARABLE_BLEND_MODE and - * PDF_NON_SEPARABLE_BLEND_MODE macros, which take the blend function as an - * argument. Note that this implementation operates on premultiplied colors, - * while the PDF specification does not. Therefore the code uses the formula - * ar.Cra = (1 – as) . Dca + (1 – ad) . Sca + B(Dca, ad, Sca, as) + * + * αr × Cr = (1 – αs) × αb × Cb + (1 – αb) × αs × Cs + αb × αs × B(Cb, Cs) + * + * with B() is the blend function. When B(Cb, Cs) = Cs, this formula + * reduces to the regular OVER operator. + * + * Cs and Cb are not premultiplied, so in our implementation we instead + * use: + * + * cr = (1 – αs) × cb + (1 – αb) × cs + αb × αs × B (cb/αb, cs/αs) + * + * where cr, cs, and cb are premultiplied colors, and where the + * + * αb × αs × B(cb/αb, cs/αs) + * + * part is first arithmetically simplified under the assumption that αb + * and αs are not 0, and then updated to produce a meaningful result when + * they are. + * + * For all the blend mode operators, the alpha channel is given by + * + * αr = αs + αb + αb × αs */ #define MAKE_SEPARABLE_PDF_COMBINERS(name) \ @@ -355,18 +376,55 @@ MAKE_PD_COMBINERS (conjoint_xor, ONE_MINUS_DA_OVER_SA, ONE_MINUS_SA_OVER_DA) \ MAKE_COMBINERS (name, combine_ ## name ## _a, combine_ ## name ## _c) +/* + * Multiply + * + * ad * as * B(d / ad, s / as) + * = ad * as * d/ad * s/as + * = d * s + * + */ static force_inline float blend_multiply (float sa, float s, float da, float d) { return d * s; } +/* + * Screen + * + * ad * as * B(d/ad, s/as) + * = ad * as * (d/ad + s/as - s/as * d/ad) + * = ad * s + as * d - s * d + */ static force_inline float blend_screen (float sa, float s, float da, float d) { return d * sa + s * da - s * d; } +/* + * Overlay + * + * ad * as * B(d/ad, s/as) + * = ad * as * Hardlight (s, d) + * = if (d / ad < 0.5) + * as * ad * Multiply (s/as, 2 * d/ad) + * else + * as * ad * Screen (s/as, 2 * d / ad - 1) + * = if (d < 0.5 * ad) + * as * ad * s/as * 2 * d /ad + * else + * as * ad * (s/as + 2 * d / ad - 1 - s / as * (2 * d / ad - 1)) + * = if (2 * d < ad) + * 2 * s * d + * else + * ad * s + 2 * as * d - as * ad - ad * s * (2 * d / ad - 1) + * = if (2 * d < ad) + * 2 * s * d + * else + * as * ad - 2 * (ad - d) * (as - s) + */ static force_inline float blend_overlay (float sa, float s, float da, float d) { @@ -376,6 +434,13 @@ blend_overlay (float sa, float s, float da, float d) return sa * da - 2 * (da - d) * (sa - s); } +/* + * Darken + * + * ad * as * B(d/ad, s/as) + * = ad * as * MIN(d/ad, s/as) + * = MIN (as * d, ad * s) + */ static force_inline float blend_darken (float sa, float s, float da, float d) { @@ -388,6 +453,13 @@ blend_darken (float sa, float s, float da, float d) return s; } +/* + * Lighten + * + * ad * as * B(d/ad, s/as) + * = ad * as * MAX(d/ad, s/as) + * = MAX (as * d, ad * s) + */ static force_inline float blend_lighten (float sa, float s, float da, float d) { @@ -400,6 +472,24 @@ blend_lighten (float sa, float s, float da, float d) return d; } +/* + * Color dodge + * + * ad * as * B(d/ad, s/as) + * = if d/ad = 0 + * ad * as * 0 + * else if (d/ad >= (1 - s/as) + * ad * as * 1 + * else + * ad * as * ((d/ad) / (1 - s/as)) + * = if d = 0 + * 0 + * elif as * d >= ad * (as - s) + * ad * as + * else + * as * (as * d / (as - s)) + * + */ static force_inline float blend_color_dodge (float sa, float s, float da, float d) { @@ -413,6 +503,26 @@ blend_color_dodge (float sa, float s, float da, float d) return sa * sa * d / (sa - s); } +/* + * Color burn + * + * We modify the first clause "if d = 1" to "if d >= 1" since with + * premultiplied colors d > 1 can actually happen. + * + * ad * as * B(d/ad, s/as) + * = if d/ad >= 1 + * ad * as * 1 + * elif (1 - d/ad) >= s/as + * ad * as * 0 + * else + * ad * as * (1 - ((1 - d/ad) / (s/as))) + * = if d >= ad + * ad * as + * elif as * ad - as * d >= ad * s + * 0 + * else + * ad * as - as * as * (ad - d) / s + */ static force_inline float blend_color_burn (float sa, float s, float da, float d) { @@ -426,6 +536,23 @@ blend_color_burn (float sa, float s, float da, float d) return sa * (da - sa * (da - d) / s); } +/* + * Hard light + * + * ad * as * B(d/ad, s/as) + * = if (s/as <= 0.5) + * ad * as * Multiply (d/ad, 2 * s/as) + * else + * ad * as * Screen (d/ad, 2 * s/as - 1) + * = if 2 * s <= as + * ad * as * d/ad * 2 * s / as + * else + * ad * as * (d/ad + (2 * s/as - 1) + d/ad * (2 * s/as - 1)) + * = if 2 * s <= as + * 2 * s * d + * else + * as * ad - 2 * (ad - d) * (as - s) + */ static force_inline float blend_hard_light (float sa, float s, float da, float d) { @@ -435,10 +562,27 @@ blend_hard_light (float sa, float s, float da, float d) return sa * da - 2 * (da - d) * (sa - s); } +/* + * Soft light + * + * ad * as * B(d/ad, s/as) + * = if (s/as <= 0.5) + * ad * as * (d/ad - (1 - 2 * s/as) * d/ad * (1 - d/ad)) + * else if (d/ad <= 0.25) + * ad * as * (d/ad + (2 * s/as - 1) * ((((16 * d/ad - 12) * d/ad + 4) * d/ad) - d/ad)) + * else + * ad * as * (d/ad + (2 * s/as - 1) * sqrt (d/ad)) + * = if (2 * s <= as) + * d * as - d * (ad - d) * (as - 2 * s) / ad; + * else if (4 * d <= ad) + * (2 * s - as) * d * ((16 * d / ad - 12) * d / ad + 3); + * else + * d * as + (sqrt (d * ad) - d) * (2 * s - as); + */ static force_inline float blend_soft_light (float sa, float s, float da, float d) { - if (2 * s < sa) + if (2 * s <= sa) { if (FLOAT_IS_ZERO (da)) return d * sa; @@ -449,7 +593,7 @@ blend_soft_light (float sa, float s, float da, float d) { if (FLOAT_IS_ZERO (da)) { - return 0.0f; + return d * sa; } else { @@ -461,6 +605,20 @@ blend_soft_light (float sa, float s, float da, float d) } } +/* + * Difference + * + * ad * as * B(s/as, d/ad) + * = ad * as * abs (s/as - d/ad) + * = if (s/as <= d/ad) + * ad * as * (d/ad - s/as) + * else + * ad * as * (s/as - d/ad) + * = if (ad * s <= as * d) + * as * d - ad * s + * else + * ad * s - as * d + */ static force_inline float blend_difference (float sa, float s, float da, float d) { @@ -473,6 +631,13 @@ blend_difference (float sa, float s, float da, float d) return sda - dsa; } +/* + * Exclusion + * + * ad * as * B(s/as, d/ad) + * = ad * as * (d/ad + s/as - 2 * d/ad * s/as) + * = as * d + ad * s - 2 * s * d + */ static force_inline float blend_exclusion (float sa, float s, float da, float d) { @@ -492,116 +657,79 @@ MAKE_SEPARABLE_PDF_COMBINERS (difference) MAKE_SEPARABLE_PDF_COMBINERS (exclusion) /* - * PDF nonseperable blend modes. - * - * These are implemented using the following functions to operate in Hsl - * space, with Cmax, Cmid, Cmin referring to the max, mid and min value - * of the red, green and blue components. + * PDF nonseperable blend modes are implemented using the following functions + * to operate in Hsl space, with Cmax, Cmid, Cmin referring to the max, mid + * and min value of the red, green and blue components. * * LUM (C) = 0.3 × Cred + 0.59 × Cgreen + 0.11 × Cblue * * clip_color (C): - * l = LUM (C) - * min = Cmin - * max = Cmax - * if n < 0.0 - * C = l + (((C – l) × l) ⁄ (l – min)) - * if x > 1.0 - * C = l + (((C – l) × (1 – l)) (max – l)) - * return C + * l = LUM (C) + * min = Cmin + * max = Cmax + * if n < 0.0 + * C = l + (((C – l) × l) ⁄ (l – min)) + * if x > 1.0 + * C = l + (((C – l) × (1 – l) ) ⁄ (max – l)) + * return C * * set_lum (C, l): - * d = l – LUM (C) - * C += d - * return clip_color (C) + * d = l – LUM (C) + * C += d + * return clip_color (C) * * SAT (C) = CH_MAX (C) - CH_MIN (C) * * set_sat (C, s): - * if Cmax > Cmin - * Cmid = ( ( ( Cmid – Cmin ) × s ) ⁄ ( Cmax – Cmin ) ) - * Cmax = s - * else - * Cmid = Cmax = 0.0 - * Cmin = 0.0 - * return C + * if Cmax > Cmin + * Cmid = ( ( ( Cmid – Cmin ) × s ) ⁄ ( Cmax – Cmin ) ) + * Cmax = s + * else + * Cmid = Cmax = 0.0 + * Cmin = 0.0 + * return C */ /* For premultiplied colors, we need to know what happens when C is * multiplied by a real number. LUM and SAT are linear: * - * LUM (r × C) = r × LUM (C) SAT (r × C) = r × SAT (C) + * LUM (r × C) = r × LUM (C) SAT (r * C) = r * SAT (C) * * If we extend clip_color with an extra argument a and change * - * if x >= 1.0 + * if x >= 1.0 * * into * - * if x >= a + * if x >= a * * then clip_color is also linear: * - * r * clip_color (C, a) = clip_color (r_c, ra); + * r * clip_color (C, a) = clip_color (r * C, r * a); * * for positive r. * * Similarly, we can extend set_lum with an extra argument that is just passed * on to clip_color: * - * r × set_lum ( C, l, a) + * r * set_lum (C, l, a) * - * = r × clip_color ( C + l - LUM (C), a) + * = r × clip_color (C + l - LUM (C), a) * - * = clip_color ( r * C + r × l - LUM (r × C), r * a) + * = clip_color (r * C + r × l - r * LUM (C), r * a) * - * = set_lum ( r * C, r * l, r * a) + * = set_lum (r * C, r * l, r * a) * * Finally, set_sat: * - * r * set_sat (C, s) = set_sat (x * C, r * s) + * r * set_sat (C, s) = set_sat (x * C, r * s) * - * The above holds for all non-zero x because they x'es in the fraction for + * The above holds for all non-zero x, because the x'es in the fraction for * C_mid cancel out. Specifically, it holds for x = r: * - * r * set_sat (C, s) = set_sat (r_c, rs) - * - * - * - * - * So, for the non-separable PDF blend modes, we have (using s, d for - * non-premultiplied colors, and S, D for premultiplied: - * - * Color: - * - * a_s * a_d * B(s, d) - * = a_s * a_d * set_lum (S/a_s, LUM (D/a_d), 1) - * = set_lum (S * a_d, a_s * LUM (D), a_s * a_d) - * - * - * Luminosity: - * - * a_s * a_d * B(s, d) - * = a_s * a_d * set_lum (D/a_d, LUM(S/a_s), 1) - * = set_lum (a_s * D, a_d * LUM(S), a_s * a_d) - * - * - * Saturation: - * - * a_s * a_d * B(s, d) - * = a_s * a_d * set_lum (set_sat (D/a_d, SAT (S/a_s)), LUM (D/a_d), 1) - * = set_lum (a_s * a_d * set_sat (D/a_d, SAT (S/a_s)), - * a_s * LUM (D), a_s * a_d) - * = set_lum (set_sat (a_s * D, a_d * SAT (S), a_s * LUM (D), a_s * a_d)) - * - * Hue: - * - * a_s * a_d * B(s, d) - * = a_s * a_d * set_lum (set_sat (S/a_s, SAT (D/a_d)), LUM (D/a_d), 1) - * = set_lum (set_sat (a_d * S, a_s * SAT (D)), a_s * LUM (D), a_s * a_d) + * r * set_sat (C, s) = set_sat (r * C, r * s) * */ - typedef struct { float r; @@ -769,9 +897,12 @@ set_sat (rgb_t *src, float sat) *min = 0.0f; } -/* - * Hue: - * B(Cb, Cs) = set_lum (set_sat (Cs, SAT (Cb)), LUM (Cb)) +/* Hue: + * + * as * ad * B(s/as, d/as) + * = as * ad * set_lum (set_sat (s/as, SAT (d/ad)), LUM (d/ad), 1) + * = set_lum (set_sat (ad * s, as * SAT (d)), as * LUM (d), as * ad) + * */ static force_inline void blend_hsl_hue (rgb_t *res, @@ -786,9 +917,14 @@ blend_hsl_hue (rgb_t *res, set_lum (res, sa * da, get_lum (dest) * sa); } -/* - * Saturation: - * B(Cb, Cs) = set_lum (set_sat (Cb, SAT (Cs)), LUM (Cb)) +/* + * Saturation + * + * as * ad * B(s/as, d/ad) + * = as * ad * set_lum (set_sat (d/ad, SAT (s/as)), LUM (d/ad), 1) + * = set_lum (as * ad * set_sat (d/ad, SAT (s/as)), + * as * LUM (d), as * ad) + * = set_lum (set_sat (as * d, ad * SAT (s), as * LUM (d), as * ad)) */ static force_inline void blend_hsl_saturation (rgb_t *res, @@ -803,9 +939,12 @@ blend_hsl_saturation (rgb_t *res, set_lum (res, sa * da, get_lum (dest) * sa); } -/* - * Color: - * B(Cb, Cs) = set_lum (Cs, LUM (Cb)) +/* + * Color + * + * as * ad * B(s/as, d/as) + * = as * ad * set_lum (s/as, LUM (d/ad), 1) + * = set_lum (s * ad, as * LUM (d), as * ad) */ static force_inline void blend_hsl_color (rgb_t *res, @@ -820,8 +959,11 @@ blend_hsl_color (rgb_t *res, } /* - * Luminosity: - * B(Cb, Cs) = set_lum (Cb, LUM (Cs)) + * Luminosity + * + * as * ad * B(s/as, d/ad) + * = as * ad * set_lum (d/ad, LUM (s/as), 1) + * = set_lum (as * d, ad * LUM (s), as * ad) */ static force_inline void blend_hsl_luminosity (rgb_t *res, diff --git a/lib/pixman/pixman/pixman-combine32.c b/lib/pixman/pixman/pixman-combine32.c index 450114a52..4c484d3e3 100644 --- a/lib/pixman/pixman/pixman-combine32.c +++ b/lib/pixman/pixman/pixman-combine32.c @@ -434,36 +434,6 @@ combine_add_u (pixman_implementation_t *imp, } } -static void -combine_saturate_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s = combine_mask (src, mask, i); - uint32_t d = *(dest + i); - uint16_t sa, da; - - sa = s >> A_SHIFT; - da = ~d >> A_SHIFT; - if (sa > da) - { - sa = DIV_UN8 (da, sa); - UN8x4_MUL_UN8 (s, sa); - } - ; - UN8x4_ADD_UN8x4 (d, s); - *(dest + i) = d; - } -} - - /* * PDF blend modes: * @@ -571,6 +541,15 @@ combine_multiply_ca (pixman_implementation_t *imp, } } +#define CLAMP(v, low, high) \ + do \ + { \ + if (v < (low)) \ + v = (low); \ + if (v > (high)) \ + v = (high); \ + } while (0) + #define PDF_SEPARABLE_BLEND_MODE(name) \ static void \ combine_ ## name ## _u (pixman_implementation_t *imp, \ @@ -589,16 +568,28 @@ combine_multiply_ca (pixman_implementation_t *imp, uint8_t isa = ~sa; \ uint8_t da = ALPHA_8 (d); \ uint8_t ida = ~da; \ - uint32_t result; \ - \ - result = d; \ - UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (result, isa, s, ida); \ + int32_t ra, rr, rg, rb; \ \ - *(dest + i) = result + \ - (DIV_ONE_UN8 (sa * (uint32_t)da) << A_SHIFT) + \ - (blend_ ## name (RED_8 (d), da, RED_8 (s), sa) << R_SHIFT) + \ - (blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), sa) << G_SHIFT) + \ - (blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), sa)); \ + ra = da * 0xff + sa * 0xff - sa * da; \ + rr = isa * RED_8 (d) + ida * RED_8 (s); \ + rg = isa * GREEN_8 (d) + ida * GREEN_8 (s); \ + rb = isa * BLUE_8 (d) + ida * BLUE_8 (s); \ + \ + rr += blend_ ## name (RED_8 (d), da, RED_8 (s), sa); \ + rg += blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), sa); \ + rb += blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), sa); \ + \ + CLAMP (ra, 0, 255 * 255); \ + CLAMP (rr, 0, 255 * 255); \ + CLAMP (rg, 0, 255 * 255); \ + CLAMP (rb, 0, 255 * 255); \ + \ + ra = DIV_ONE_UN8 (ra); \ + rr = DIV_ONE_UN8 (rr); \ + rg = DIV_ONE_UN8 (rg); \ + rb = DIV_ONE_UN8 (rb); \ + \ + *(dest + i) = ra << 24 | rr << 16 | rg << 8 | rb; \ } \ } \ \ @@ -618,20 +609,35 @@ combine_multiply_ca (pixman_implementation_t *imp, uint32_t d = *(dest + i); \ uint8_t da = ALPHA_8 (d); \ uint8_t ida = ~da; \ - uint32_t result; \ - \ + int32_t ra, rr, rg, rb; \ + uint8_t ira, iga, iba; \ + \ combine_mask_ca (&s, &m); \ - \ - result = d; \ - UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (result, ~m, s, ida); \ - \ - result += \ - (DIV_ONE_UN8 (ALPHA_8 (m) * (uint32_t)da) << A_SHIFT) + \ - (blend_ ## name (RED_8 (d), da, RED_8 (s), RED_8 (m)) << R_SHIFT) + \ - (blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), GREEN_8 (m)) << G_SHIFT) + \ - (blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), BLUE_8 (m))); \ \ - *(dest + i) = result; \ + ira = ~RED_8 (m); \ + iga = ~GREEN_8 (m); \ + iba = ~BLUE_8 (m); \ + \ + ra = da * 0xff + ALPHA_8 (s) * 0xff - ALPHA_8 (s) * da; \ + rr = ira * RED_8 (d) + ida * RED_8 (s); \ + rg = iga * GREEN_8 (d) + ida * GREEN_8 (s); \ + rb = iba * BLUE_8 (d) + ida * BLUE_8 (s); \ + \ + rr += blend_ ## name (RED_8 (d), da, RED_8 (s), RED_8 (m)); \ + rg += blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), GREEN_8 (m)); \ + rb += blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), BLUE_8 (m)); \ + \ + CLAMP (ra, 0, 255 * 255); \ + CLAMP (rr, 0, 255 * 255); \ + CLAMP (rg, 0, 255 * 255); \ + CLAMP (rb, 0, 255 * 255); \ + \ + ra = DIV_ONE_UN8 (ra); \ + rr = DIV_ONE_UN8 (rr); \ + rg = DIV_ONE_UN8 (rg); \ + rb = DIV_ONE_UN8 (rb); \ + \ + *(dest + i) = ra << 24 | rr << 16 | rg << 8 | rb; \ } \ } @@ -642,10 +648,10 @@ combine_multiply_ca (pixman_implementation_t *imp, * = ad * as * (d/ad + s/as - s/as * d/ad) * = ad * s + as * d - s * d */ -static inline uint32_t -blend_screen (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) +static inline int32_t +blend_screen (int32_t d, int32_t ad, int32_t s, int32_t as) { - return DIV_ONE_UN8 (s * ad + d * as - s * d); + return s * ad + d * as - s * d; } PDF_SEPARABLE_BLEND_MODE (screen) @@ -672,8 +678,8 @@ PDF_SEPARABLE_BLEND_MODE (screen) * else * as * ad - 2 * (ad - d) * (as - s) */ -static inline uint32_t -blend_overlay (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) +static inline int32_t +blend_overlay (int32_t d, int32_t ad, int32_t s, int32_t as) { uint32_t r; @@ -682,7 +688,7 @@ blend_overlay (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) else r = as * ad - 2 * (ad - d) * (as - s); - return DIV_ONE_UN8 (r); + return r; } PDF_SEPARABLE_BLEND_MODE (overlay) @@ -694,1013 +700,111 @@ PDF_SEPARABLE_BLEND_MODE (overlay) * = ad * as * MIN(d/ad, s/as) * = MIN (as * d, ad * s) */ -static inline uint32_t -blend_darken (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - s = ad * s; - d = as * d; - - return DIV_ONE_UN8 (s > d ? d : s); -} - -PDF_SEPARABLE_BLEND_MODE (darken) - -/* - * Lighten - * - * ad * as * B(d/ad, s/as) - * = ad * as * MAX(d/ad, s/as) - * = MAX (as * d, ad * s) - */ -static inline uint32_t -blend_lighten (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) +static inline int32_t +blend_darken (int32_t d, int32_t ad, int32_t s, int32_t as) { s = ad * s; d = as * d; - - return DIV_ONE_UN8 (s > d ? s : d); -} - -PDF_SEPARABLE_BLEND_MODE (lighten) - -/* - * Color dodge - * - * ad * as * B(d/ad, s/as) - * = if d/ad = 0 - * ad * as * 0 - * else if (d/ad >= (1 - s/as) - * ad * as * 1 - * else - * ad * as * ((d/ad) / (1 - s/as)) - * = if d = 0 - * 0 - * elif as * d >= ad * (as - s) - * ad * as - * else - * as * (as * d / (as - s)) - * - */ -static inline uint32_t -blend_color_dodge (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - if (d == 0) - return 0; - else if (as * d >= ad * (as - s)) - return DIV_ONE_UN8 (as * ad); - else if (as - s == 0) - return DIV_ONE_UN8 (as * ad); - else - return DIV_ONE_UN8 (as * ((d * as) / ((as - s)))); -} - -PDF_SEPARABLE_BLEND_MODE (color_dodge) - -/* - * Color burn - * - * We modify the first clause "if d = 1" to "if d >= 1" since with - * premultiplied colors d > 1 can actually happen. - * - * ad * as * B(d/ad, s/as) - * = if d/ad >= 1 - * ad * as * 1 - * elif (1 - d/ad) >= s/as - * ad * as * 0 - * else - * ad * as * (1 - ((1 - d/ad) / (s/as))) - * = if d >= ad - * ad * as - * elif as * ad - as * d >= ad * s - * 0 - * else - * ad * as - as * as * (ad - d) / s - */ -static inline uint32_t -blend_color_burn (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - if (d >= ad) - return DIV_ONE_UN8 (ad * as); - else if (as * ad - as * d >= ad * s) - return 0; - else if (s == 0) - return 0; - else - return DIV_ONE_UN8 (ad * as - (as * as * (ad - d)) / s); -} - -PDF_SEPARABLE_BLEND_MODE (color_burn) - -/* - * Hard light - * - * ad * as * B(d/ad, s/as) - * = if (s/as <= 0.5) - * ad * as * Multiply (d/ad, 2 * s/as) - * else - * ad * as * Screen (d/ad, 2 * s/as - 1) - * = if 2 * s <= as - * ad * as * d/ad * 2 * s / as - * else - * ad * as * (d/ad + (2 * s/as - 1) + d/ad * (2 * s/as - 1)) - * = if 2 * s <= as - * 2 * s * d - * else - * as * ad - 2 * (ad - d) * (as - s) - */ -static inline uint32_t -blend_hard_light (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - if (2 * s < as) - return DIV_ONE_UN8 (2 * s * d); - else - return DIV_ONE_UN8 (as * ad - 2 * (ad - d) * (as - s)); -} - -PDF_SEPARABLE_BLEND_MODE (hard_light) - -/* - * Soft light - * - * ad * as * B(d/ad, s/as) - * = if (s/as <= 0.5) - * ad * as * (d/ad - (1 - 2 * s/as) * d/ad * (1 - d/ad)) - * else if (d/ad <= 0.25) - * ad * as * (d/ad + (2 * s/as - 1) * ((((16 * d/ad - 12) * d/ad + 4) * d/ad) - d/ad)) - * else - * ad * as * (d/ad + (2 * s/as - 1) * sqrt (d/ad)) - * = if (2 * s <= as) - * d * as - d * (ad - d) * (as - 2 * s) / ad; - * else if (4 * d <= ad) - * (2 * s - as) * d * ((16 * d / ad - 12) * d / ad + 3); - * else - * d * as + (sqrt (d * ad) - d) * (2 * s - as); - */ -static inline uint32_t -blend_soft_light (uint32_t d_org, - uint32_t ad_org, - uint32_t s_org, - uint32_t as_org) -{ - double d = d_org * (1.0 / MASK); - double ad = ad_org * (1.0 / MASK); - double s = s_org * (1.0 / MASK); - double as = as_org * (1.0 / MASK); - double r; - - if (2 * s < as) - { - if (ad == 0) - r = d * as; - else - r = d * as - d * (ad - d) * (as - 2 * s) / ad; - } - else if (ad == 0) - { - r = 0; - } - else if (4 * d <= ad) - { - r = d * as + - (2 * s - as) * d * ((16 * d / ad - 12) * d / ad + 3); - } - else - { - r = d * as + (sqrt (d * ad) - d) * (2 * s - as); - } - return r * MASK + 0.5; -} - -PDF_SEPARABLE_BLEND_MODE (soft_light) - -/* - * Difference - * - * ad * as * B(s/as, d/ad) - * = ad * as * abs (s/as - d/ad) - * = if (s/as <= d/ad) - * ad * as * (d/ad - s/as) - * else - * ad * as * (s/as - d/ad) - * = if (ad * s <= as * d) - * as * d - ad * s - * else - * ad * s - as * d - */ -static inline uint32_t -blend_difference (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - uint32_t das = d * as; - uint32_t sad = s * ad; - - if (sad < das) - return DIV_ONE_UN8 (das - sad); - else - return DIV_ONE_UN8 (sad - das); -} - -PDF_SEPARABLE_BLEND_MODE (difference) - -/* - * Exclusion - * - * ad * as * B(s/as, d/ad) - * = ad * as * (d/ad + s/as - 2 * d/ad * s/as) - * = as * d + ad * s - 2 * s * d - */ - -/* This can be made faster by writing it directly and not using - * PDF_SEPARABLE_BLEND_MODE, but that's a performance optimization */ - -static inline uint32_t -blend_exclusion (uint32_t d, uint32_t ad, uint32_t s, uint32_t as) -{ - return DIV_ONE_UN8 (s * ad + d * as - 2 * d * s); -} - -PDF_SEPARABLE_BLEND_MODE (exclusion) - -#undef PDF_SEPARABLE_BLEND_MODE - -/* - * PDF nonseperable blend modes are implemented using the following functions - * to operate in Hsl space, with Cmax, Cmid, Cmin referring to the max, mid - * and min value of the red, green and blue components. - * - * LUM (C) = 0.3 × Cred + 0.59 × Cgreen + 0.11 × Cblue - * - * clip_color (C): - * l = LUM (C) - * min = Cmin - * max = Cmax - * if n < 0.0 - * C = l + (((C – l) × l) ⁄ (l – min)) - * if x > 1.0 - * C = l + (((C – l) × (1 – l) ) ⁄ (max – l)) - * return C - * - * set_lum (C, l): - * d = l – LUM (C) - * C += d - * return clip_color (C) - * - * SAT (C) = CH_MAX (C) - CH_MIN (C) - * - * set_sat (C, s): - * if Cmax > Cmin - * Cmid = ( ( ( Cmid – Cmin ) × s ) ⁄ ( Cmax – Cmin ) ) - * Cmax = s - * else - * Cmid = Cmax = 0.0 - * Cmin = 0.0 - * return C - */ - -/* For premultiplied colors, we need to know what happens when C is - * multiplied by a real number. LUM and SAT are linear: - * - * LUM (r × C) = r × LUM (C) SAT (r * C) = r * SAT (C) - * - * If we extend clip_color with an extra argument a and change - * - * if x >= 1.0 - * - * into - * - * if x >= a - * - * then clip_color is also linear: - * - * r * clip_color (C, a) = clip_color (r * C, r * a); - * - * for positive r. - * - * Similarly, we can extend set_lum with an extra argument that is just passed - * on to clip_color: - * - * r * set_lum (C, l, a) - * - * = r × clip_color (C + l - LUM (C), a) - * - * = clip_color (r * C + r × l - r * LUM (C), r * a) - * - * = set_lum (r * C, r * l, r * a) - * - * Finally, set_sat: - * - * r * set_sat (C, s) = set_sat (x * C, r * s) - * - * The above holds for all non-zero x, because the x'es in the fraction for - * C_mid cancel out. Specifically, it holds for x = r: - * - * r * set_sat (C, s) = set_sat (r * C, r * s) - * - */ - -#define CH_MIN(c) (c[0] < c[1] ? (c[0] < c[2] ? c[0] : c[2]) : (c[1] < c[2] ? c[1] : c[2])) -#define CH_MAX(c) (c[0] > c[1] ? (c[0] > c[2] ? c[0] : c[2]) : (c[1] > c[2] ? c[1] : c[2])) -#define LUM(c) ((c[0] * 30 + c[1] * 59 + c[2] * 11) / 100) -#define SAT(c) (CH_MAX (c) - CH_MIN (c)) - -#define PDF_NON_SEPARABLE_BLEND_MODE(name) \ - static void \ - combine_ ## name ## _u (pixman_implementation_t *imp, \ - pixman_op_t op, \ - uint32_t * dest, \ - const uint32_t * src, \ - const uint32_t * mask, \ - int width) \ - { \ - int i; \ - for (i = 0; i < width; ++i) \ - { \ - uint32_t s = combine_mask (src, mask, i); \ - uint32_t d = *(dest + i); \ - uint8_t sa = ALPHA_8 (s); \ - uint8_t isa = ~sa; \ - uint8_t da = ALPHA_8 (d); \ - uint8_t ida = ~da; \ - uint32_t result; \ - uint32_t sc[3], dc[3], c[3]; \ - \ - result = d; \ - UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (result, isa, s, ida); \ - dc[0] = RED_8 (d); \ - sc[0] = RED_8 (s); \ - dc[1] = GREEN_8 (d); \ - sc[1] = GREEN_8 (s); \ - dc[2] = BLUE_8 (d); \ - sc[2] = BLUE_8 (s); \ - blend_ ## name (c, dc, da, sc, sa); \ - \ - *(dest + i) = result + \ - (DIV_ONE_UN8 (sa * (uint32_t)da) << A_SHIFT) + \ - (DIV_ONE_UN8 (c[0]) << R_SHIFT) + \ - (DIV_ONE_UN8 (c[1]) << G_SHIFT) + \ - (DIV_ONE_UN8 (c[2])); \ - } \ - } - -static void -set_lum (uint32_t dest[3], uint32_t src[3], uint32_t sa, uint32_t lum) -{ - double a, l, min, max; - double tmp[3]; - - a = sa * (1.0 / MASK); - - l = lum * (1.0 / MASK); - tmp[0] = src[0] * (1.0 / MASK); - tmp[1] = src[1] * (1.0 / MASK); - tmp[2] = src[2] * (1.0 / MASK); - - l = l - LUM (tmp); - tmp[0] += l; - tmp[1] += l; - tmp[2] += l; - - /* clip_color */ - l = LUM (tmp); - min = CH_MIN (tmp); - max = CH_MAX (tmp); - - if (min < 0) - { - if (l - min == 0.0) - { - tmp[0] = 0; - tmp[1] = 0; - tmp[2] = 0; - } - else - { - tmp[0] = l + (tmp[0] - l) * l / (l - min); - tmp[1] = l + (tmp[1] - l) * l / (l - min); - tmp[2] = l + (tmp[2] - l) * l / (l - min); - } - } - if (max > a) - { - if (max - l == 0.0) - { - tmp[0] = a; - tmp[1] = a; - tmp[2] = a; - } - else - { - tmp[0] = l + (tmp[0] - l) * (a - l) / (max - l); - tmp[1] = l + (tmp[1] - l) * (a - l) / (max - l); - tmp[2] = l + (tmp[2] - l) * (a - l) / (max - l); - } - } - - dest[0] = tmp[0] * MASK + 0.5; - dest[1] = tmp[1] * MASK + 0.5; - dest[2] = tmp[2] * MASK + 0.5; -} - -static void -set_sat (uint32_t dest[3], uint32_t src[3], uint32_t sat) -{ - int id[3]; - uint32_t min, max; - - if (src[0] > src[1]) - { - if (src[0] > src[2]) - { - id[0] = 0; - if (src[1] > src[2]) - { - id[1] = 1; - id[2] = 2; - } - else - { - id[1] = 2; - id[2] = 1; - } - } - else - { - id[0] = 2; - id[1] = 0; - id[2] = 1; - } - } - else - { - if (src[0] > src[2]) - { - id[0] = 1; - id[1] = 0; - id[2] = 2; - } - else - { - id[2] = 0; - if (src[1] > src[2]) - { - id[0] = 1; - id[1] = 2; - } - else - { - id[0] = 2; - id[1] = 1; - } - } - } - - max = dest[id[0]]; - min = dest[id[2]]; - if (max > min) - { - dest[id[1]] = (dest[id[1]] - min) * sat / (max - min); - dest[id[0]] = sat; - dest[id[2]] = 0; - } - else - { - dest[0] = dest[1] = dest[2] = 0; - } -} - -/* Hue: - * - * as * ad * B(s/as, d/as) - * = as * ad * set_lum (set_sat (s/as, SAT (d/ad)), LUM (d/ad), 1) - * = set_lum (set_sat (ad * s, as * SAT (d)), as * LUM (d), as * ad) - * - */ -static inline void -blend_hsl_hue (uint32_t r[3], - uint32_t d[3], - uint32_t ad, - uint32_t s[3], - uint32_t as) -{ - r[0] = s[0] * ad; - r[1] = s[1] * ad; - r[2] = s[2] * ad; - set_sat (r, r, SAT (d) * as); - set_lum (r, r, as * ad, LUM (d) * as); -} - -PDF_NON_SEPARABLE_BLEND_MODE (hsl_hue) - -/* - * Saturation - * - * as * ad * B(s/as, d/ad) - * = as * ad * set_lum (set_sat (d/ad, SAT (s/as)), LUM (d/ad), 1) - * = set_lum (as * ad * set_sat (d/ad, SAT (s/as)), - * as * LUM (d), as * ad) - * = set_lum (set_sat (as * d, ad * SAT (s), as * LUM (d), as * ad)) - */ -static inline void -blend_hsl_saturation (uint32_t r[3], - uint32_t d[3], - uint32_t ad, - uint32_t s[3], - uint32_t as) -{ - r[0] = d[0] * as; - r[1] = d[1] * as; - r[2] = d[2] * as; - set_sat (r, r, SAT (s) * ad); - set_lum (r, r, as * ad, LUM (d) * as); -} - -PDF_NON_SEPARABLE_BLEND_MODE (hsl_saturation) - -/* - * Color - * - * as * ad * B(s/as, d/as) - * = as * ad * set_lum (s/as, LUM (d/ad), 1) - * = set_lum (s * ad, as * LUM (d), as * ad) - */ -static inline void -blend_hsl_color (uint32_t r[3], - uint32_t d[3], - uint32_t ad, - uint32_t s[3], - uint32_t as) -{ - r[0] = s[0] * ad; - r[1] = s[1] * ad; - r[2] = s[2] * ad; - set_lum (r, r, as * ad, LUM (d) * as); -} - -PDF_NON_SEPARABLE_BLEND_MODE (hsl_color) - -/* - * Luminosity - * - * as * ad * B(s/as, d/ad) - * = as * ad * set_lum (d/ad, LUM (s/as), 1) - * = set_lum (as * d, ad * LUM (s), as * ad) - */ -static inline void -blend_hsl_luminosity (uint32_t r[3], - uint32_t d[3], - uint32_t ad, - uint32_t s[3], - uint32_t as) -{ - r[0] = d[0] * as; - r[1] = d[1] * as; - r[2] = d[2] * as; - set_lum (r, r, as * ad, LUM (s) * ad); -} - -PDF_NON_SEPARABLE_BLEND_MODE (hsl_luminosity) - -#undef SAT -#undef LUM -#undef CH_MAX -#undef CH_MIN -#undef PDF_NON_SEPARABLE_BLEND_MODE - -/* All of the disjoint/conjoint composing functions - * - * The four entries in the first column indicate what source contributions - * come from each of the four areas of the picture -- areas covered by neither - * A nor B, areas covered only by A, areas covered only by B and finally - * areas covered by both A and B. - * - * Disjoint Conjoint - * Fa Fb Fa Fb - * (0,0,0,0) 0 0 0 0 - * (0,A,0,A) 1 0 1 0 - * (0,0,B,B) 0 1 0 1 - * (0,A,B,A) 1 min((1-a)/b,1) 1 max(1-a/b,0) - * (0,A,B,B) min((1-b)/a,1) 1 max(1-b/a,0) 1 - * (0,0,0,A) max(1-(1-b)/a,0) 0 min(1,b/a) 0 - * (0,0,0,B) 0 max(1-(1-a)/b,0) 0 min(a/b,1) - * (0,A,0,0) min(1,(1-b)/a) 0 max(1-b/a,0) 0 - * (0,0,B,0) 0 min(1,(1-a)/b) 0 max(1-a/b,0) - * (0,0,B,A) max(1-(1-b)/a,0) min(1,(1-a)/b) min(1,b/a) max(1-a/b,0) - * (0,A,0,B) min(1,(1-b)/a) max(1-(1-a)/b,0) max(1-b/a,0) min(1,a/b) - * (0,A,B,0) min(1,(1-b)/a) min(1,(1-a)/b) max(1-b/a,0) max(1-a/b,0) - * - * See http://marc.info/?l=xfree-render&m=99792000027857&w=2 for more - * information about these operators. - */ - -#define COMBINE_A_OUT 1 -#define COMBINE_A_IN 2 -#define COMBINE_B_OUT 4 -#define COMBINE_B_IN 8 - -#define COMBINE_CLEAR 0 -#define COMBINE_A (COMBINE_A_OUT | COMBINE_A_IN) -#define COMBINE_B (COMBINE_B_OUT | COMBINE_B_IN) -#define COMBINE_A_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_A_IN) -#define COMBINE_B_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_B_IN) -#define COMBINE_A_ATOP (COMBINE_B_OUT | COMBINE_A_IN) -#define COMBINE_B_ATOP (COMBINE_A_OUT | COMBINE_B_IN) -#define COMBINE_XOR (COMBINE_A_OUT | COMBINE_B_OUT) - -/* portion covered by a but not b */ -static uint8_t -combine_disjoint_out_part (uint8_t a, uint8_t b) -{ - /* min (1, (1-b) / a) */ - - b = ~b; /* 1 - b */ - if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ - return MASK; /* 1 */ - return DIV_UN8 (b, a); /* (1-b) / a */ -} - -/* portion covered by both a and b */ -static uint8_t -combine_disjoint_in_part (uint8_t a, uint8_t b) -{ - /* max (1-(1-b)/a,0) */ - /* = - min ((1-b)/a - 1, 0) */ - /* = 1 - min (1, (1-b)/a) */ - - b = ~b; /* 1 - b */ - if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ - return 0; /* 1 - 1 */ - return ~DIV_UN8(b, a); /* 1 - (1-b) / a */ -} - -/* portion covered by a but not b */ -static uint8_t -combine_conjoint_out_part (uint8_t a, uint8_t b) -{ - /* max (1-b/a,0) */ - /* = 1-min(b/a,1) */ - - /* min (1, (1-b) / a) */ - - if (b >= a) /* b >= a -> b/a >= 1 */ - return 0x00; /* 0 */ - return ~DIV_UN8(b, a); /* 1 - b/a */ -} - -/* portion covered by both a and b */ -static uint8_t -combine_conjoint_in_part (uint8_t a, uint8_t b) -{ - /* min (1,b/a) */ - - if (b >= a) /* b >= a -> b/a >= 1 */ - return MASK; /* 1 */ - return DIV_UN8 (b, a); /* b/a */ -} - -#define GET_COMP(v, i) ((uint16_t) (uint8_t) ((v) >> i)) - -#define ADD(x, y, i, t) \ - ((t) = GET_COMP (x, i) + GET_COMP (y, i), \ - (uint32_t) ((uint8_t) ((t) | (0 - ((t) >> G_SHIFT)))) << (i)) - -#define GENERIC(x, y, i, ax, ay, t, u, v) \ - ((t) = (MUL_UN8 (GET_COMP (y, i), ay, (u)) + \ - MUL_UN8 (GET_COMP (x, i), ax, (v))), \ - (uint32_t) ((uint8_t) ((t) | \ - (0 - ((t) >> G_SHIFT)))) << (i)) - -static void -combine_disjoint_general_u (uint32_t * dest, - const uint32_t *src, - const uint32_t *mask, - int width, - uint8_t combine) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s = combine_mask (src, mask, i); - uint32_t d = *(dest + i); - uint32_t m, n, o, p; - uint16_t Fa, Fb, t, u, v; - uint8_t sa = s >> A_SHIFT; - uint8_t da = d >> A_SHIFT; - - switch (combine & COMBINE_A) - { - default: - Fa = 0; - break; - - case COMBINE_A_OUT: - Fa = combine_disjoint_out_part (sa, da); - break; - - case COMBINE_A_IN: - Fa = combine_disjoint_in_part (sa, da); - break; - - case COMBINE_A: - Fa = MASK; - break; - } - - switch (combine & COMBINE_B) - { - default: - Fb = 0; - break; - - case COMBINE_B_OUT: - Fb = combine_disjoint_out_part (da, sa); - break; - - case COMBINE_B_IN: - Fb = combine_disjoint_in_part (da, sa); - break; - - case COMBINE_B: - Fb = MASK; - break; - } - m = GENERIC (s, d, 0, Fa, Fb, t, u, v); - n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); - o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); - p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); - s = m | n | o | p; - *(dest + i) = s; - } -} - -static void -combine_disjoint_over_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s = combine_mask (src, mask, i); - uint16_t a = s >> A_SHIFT; - - if (s != 0x00) - { - uint32_t d = *(dest + i); - a = combine_disjoint_out_part (d >> A_SHIFT, a); - UN8x4_MUL_UN8_ADD_UN8x4 (d, a, s); - - *(dest + i) = d; - } - } -} - -static void -combine_disjoint_in_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_IN); -} - -static void -combine_disjoint_in_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_IN); -} - -static void -combine_disjoint_out_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); -} - -static void -combine_disjoint_out_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); -} - -static void -combine_disjoint_atop_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); -} - -static void -combine_disjoint_atop_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); -} - -static void -combine_disjoint_xor_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_u (dest, src, mask, width, COMBINE_XOR); -} - -static void -combine_conjoint_general_u (uint32_t * dest, - const uint32_t *src, - const uint32_t *mask, - int width, - uint8_t combine) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s = combine_mask (src, mask, i); - uint32_t d = *(dest + i); - uint32_t m, n, o, p; - uint16_t Fa, Fb, t, u, v; - uint8_t sa = s >> A_SHIFT; - uint8_t da = d >> A_SHIFT; - - switch (combine & COMBINE_A) - { - default: - Fa = 0; - break; - - case COMBINE_A_OUT: - Fa = combine_conjoint_out_part (sa, da); - break; - - case COMBINE_A_IN: - Fa = combine_conjoint_in_part (sa, da); - break; - - case COMBINE_A: - Fa = MASK; - break; - } - - switch (combine & COMBINE_B) - { - default: - Fb = 0; - break; - - case COMBINE_B_OUT: - Fb = combine_conjoint_out_part (da, sa); - break; - - case COMBINE_B_IN: - Fb = combine_conjoint_in_part (da, sa); - break; - - case COMBINE_B: - Fb = MASK; - break; - } - - m = GENERIC (s, d, 0, Fa, Fb, t, u, v); - n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); - o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); - p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); - - s = m | n | o | p; - - *(dest + i) = s; - } -} - -static void -combine_conjoint_over_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OVER); -} -static void -combine_conjoint_over_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OVER); + return s > d ? d : s; } -static void -combine_conjoint_in_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_IN); -} +PDF_SEPARABLE_BLEND_MODE (darken) -static void -combine_conjoint_in_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) +/* + * Lighten + * + * ad * as * B(d/ad, s/as) + * = ad * as * MAX(d/ad, s/as) + * = MAX (as * d, ad * s) + */ +static inline int32_t +blend_lighten (int32_t d, int32_t ad, int32_t s, int32_t as) { - combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_IN); + s = ad * s; + d = as * d; + + return s > d ? s : d; } -static void -combine_conjoint_out_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); -} +PDF_SEPARABLE_BLEND_MODE (lighten) -static void -combine_conjoint_out_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) +/* + * Hard light + * + * ad * as * B(d/ad, s/as) + * = if (s/as <= 0.5) + * ad * as * Multiply (d/ad, 2 * s/as) + * else + * ad * as * Screen (d/ad, 2 * s/as - 1) + * = if 2 * s <= as + * ad * as * d/ad * 2 * s / as + * else + * ad * as * (d/ad + (2 * s/as - 1) + d/ad * (2 * s/as - 1)) + * = if 2 * s <= as + * 2 * s * d + * else + * as * ad - 2 * (ad - d) * (as - s) + */ +static inline int32_t +blend_hard_light (int32_t d, int32_t ad, int32_t s, int32_t as) { - combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); + if (2 * s < as) + return 2 * s * d; + else + return as * ad - 2 * (ad - d) * (as - s); } -static void -combine_conjoint_atop_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); -} +PDF_SEPARABLE_BLEND_MODE (hard_light) -static void -combine_conjoint_atop_reverse_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) +/* + * Difference + * + * ad * as * B(s/as, d/ad) + * = ad * as * abs (s/as - d/ad) + * = if (s/as <= d/ad) + * ad * as * (d/ad - s/as) + * else + * ad * as * (s/as - d/ad) + * = if (ad * s <= as * d) + * as * d - ad * s + * else + * ad * s - as * d + */ +static inline int32_t +blend_difference (int32_t d, int32_t ad, int32_t s, int32_t as) { - combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); + int32_t das = d * as; + int32_t sad = s * ad; + + if (sad < das) + return das - sad; + else + return sad - das; } -static void -combine_conjoint_xor_u (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) +PDF_SEPARABLE_BLEND_MODE (difference) + +/* + * Exclusion + * + * ad * as * B(s/as, d/ad) + * = ad * as * (d/ad + s/as - 2 * d/ad * s/as) + * = as * d + ad * s - 2 * s * d + */ + +/* This can be made faster by writing it directly and not using + * PDF_SEPARABLE_BLEND_MODE, but that's a performance optimization */ + +static inline int32_t +blend_exclusion (int32_t d, int32_t ad, int32_t s, int32_t as) { - combine_conjoint_general_u (dest, src, mask, width, COMBINE_XOR); + return s * ad + d * as - 2 * d * s; } +PDF_SEPARABLE_BLEND_MODE (exclusion) + +#undef PDF_SEPARABLE_BLEND_MODE /* Component alpha combiners */ @@ -2032,428 +1136,6 @@ combine_add_ca (pixman_implementation_t *imp, } } -static void -combine_saturate_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s, d; - uint16_t sa, sr, sg, sb, da; - uint16_t t, u, v; - uint32_t m, n, o, p; - - d = *(dest + i); - s = *(src + i); - m = *(mask + i); - - combine_mask_ca (&s, &m); - - sa = (m >> A_SHIFT); - sr = (m >> R_SHIFT) & MASK; - sg = (m >> G_SHIFT) & MASK; - sb = m & MASK; - da = ~d >> A_SHIFT; - - if (sb <= da) - m = ADD (s, d, 0, t); - else - m = GENERIC (s, d, 0, (da << G_SHIFT) / sb, MASK, t, u, v); - - if (sg <= da) - n = ADD (s, d, G_SHIFT, t); - else - n = GENERIC (s, d, G_SHIFT, (da << G_SHIFT) / sg, MASK, t, u, v); - - if (sr <= da) - o = ADD (s, d, R_SHIFT, t); - else - o = GENERIC (s, d, R_SHIFT, (da << G_SHIFT) / sr, MASK, t, u, v); - - if (sa <= da) - p = ADD (s, d, A_SHIFT, t); - else - p = GENERIC (s, d, A_SHIFT, (da << G_SHIFT) / sa, MASK, t, u, v); - - *(dest + i) = m | n | o | p; - } -} - -static void -combine_disjoint_general_ca (uint32_t * dest, - const uint32_t *src, - const uint32_t *mask, - int width, - uint8_t combine) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s, d; - uint32_t m, n, o, p; - uint32_t Fa, Fb; - uint16_t t, u, v; - uint32_t sa; - uint8_t da; - - s = *(src + i); - m = *(mask + i); - d = *(dest + i); - da = d >> A_SHIFT; - - combine_mask_ca (&s, &m); - - sa = m; - - switch (combine & COMBINE_A) - { - default: - Fa = 0; - break; - - case COMBINE_A_OUT: - m = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> 0), da); - n = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; - o = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; - p = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; - Fa = m | n | o | p; - break; - - case COMBINE_A_IN: - m = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> 0), da); - n = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; - o = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; - p = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; - Fa = m | n | o | p; - break; - - case COMBINE_A: - Fa = ~0; - break; - } - - switch (combine & COMBINE_B) - { - default: - Fb = 0; - break; - - case COMBINE_B_OUT: - m = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> 0)); - n = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; - o = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; - p = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; - Fb = m | n | o | p; - break; - - case COMBINE_B_IN: - m = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> 0)); - n = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; - o = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; - p = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; - Fb = m | n | o | p; - break; - - case COMBINE_B: - Fb = ~0; - break; - } - m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); - n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); - o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); - p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); - - s = m | n | o | p; - - *(dest + i) = s; - } -} - -static void -combine_disjoint_over_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); -} - -static void -combine_disjoint_in_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); -} - -static void -combine_disjoint_in_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); -} - -static void -combine_disjoint_out_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); -} - -static void -combine_disjoint_out_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); -} - -static void -combine_disjoint_atop_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); -} - -static void -combine_disjoint_atop_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); -} - -static void -combine_disjoint_xor_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_disjoint_general_ca (dest, src, mask, width, COMBINE_XOR); -} - -static void -combine_conjoint_general_ca (uint32_t * dest, - const uint32_t *src, - const uint32_t *mask, - int width, - uint8_t combine) -{ - int i; - - for (i = 0; i < width; ++i) - { - uint32_t s, d; - uint32_t m, n, o, p; - uint32_t Fa, Fb; - uint16_t t, u, v; - uint32_t sa; - uint8_t da; - - s = *(src + i); - m = *(mask + i); - d = *(dest + i); - da = d >> A_SHIFT; - - combine_mask_ca (&s, &m); - - sa = m; - - switch (combine & COMBINE_A) - { - default: - Fa = 0; - break; - - case COMBINE_A_OUT: - m = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> 0), da); - n = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; - o = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; - p = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; - Fa = m | n | o | p; - break; - - case COMBINE_A_IN: - m = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> 0), da); - n = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; - o = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; - p = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; - Fa = m | n | o | p; - break; - - case COMBINE_A: - Fa = ~0; - break; - } - - switch (combine & COMBINE_B) - { - default: - Fb = 0; - break; - - case COMBINE_B_OUT: - m = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> 0)); - n = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; - o = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; - p = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; - Fb = m | n | o | p; - break; - - case COMBINE_B_IN: - m = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> 0)); - n = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; - o = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; - p = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; - Fb = m | n | o | p; - break; - - case COMBINE_B: - Fb = ~0; - break; - } - m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); - n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); - o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); - p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); - - s = m | n | o | p; - - *(dest + i) = s; - } -} - -static void -combine_conjoint_over_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); -} - -static void -combine_conjoint_over_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OVER); -} - -static void -combine_conjoint_in_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); -} - -static void -combine_conjoint_in_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); -} - -static void -combine_conjoint_out_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); -} - -static void -combine_conjoint_out_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); -} - -static void -combine_conjoint_atop_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); -} - -static void -combine_conjoint_atop_reverse_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); -} - -static void -combine_conjoint_xor_ca (pixman_implementation_t *imp, - pixman_op_t op, - uint32_t * dest, - const uint32_t * src, - const uint32_t * mask, - int width) -{ - combine_conjoint_general_ca (dest, src, mask, width, COMBINE_XOR); -} - void _pixman_setup_combiner_functions_32 (pixman_implementation_t *imp) { @@ -2471,51 +1153,15 @@ _pixman_setup_combiner_functions_32 (pixman_implementation_t *imp) imp->combine_32[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_u; imp->combine_32[PIXMAN_OP_XOR] = combine_xor_u; imp->combine_32[PIXMAN_OP_ADD] = combine_add_u; - imp->combine_32[PIXMAN_OP_SATURATE] = combine_saturate_u; - - /* Disjoint, unified */ - imp->combine_32[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear; - imp->combine_32[PIXMAN_OP_DISJOINT_SRC] = combine_src_u; - imp->combine_32[PIXMAN_OP_DISJOINT_DST] = combine_dst; - imp->combine_32[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_u; - imp->combine_32[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_u; - imp->combine_32[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_u; - imp->combine_32[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_u; - imp->combine_32[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_u; - imp->combine_32[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_u; - imp->combine_32[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_u; - imp->combine_32[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_u; - imp->combine_32[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_u; - - /* Conjoint, unified */ - imp->combine_32[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear; - imp->combine_32[PIXMAN_OP_CONJOINT_SRC] = combine_src_u; - imp->combine_32[PIXMAN_OP_CONJOINT_DST] = combine_dst; - imp->combine_32[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_u; - imp->combine_32[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_u; - imp->combine_32[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_u; - imp->combine_32[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_u; - imp->combine_32[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_u; - imp->combine_32[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_u; - imp->combine_32[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_u; - imp->combine_32[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_u; - imp->combine_32[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_u; imp->combine_32[PIXMAN_OP_MULTIPLY] = combine_multiply_u; imp->combine_32[PIXMAN_OP_SCREEN] = combine_screen_u; imp->combine_32[PIXMAN_OP_OVERLAY] = combine_overlay_u; imp->combine_32[PIXMAN_OP_DARKEN] = combine_darken_u; imp->combine_32[PIXMAN_OP_LIGHTEN] = combine_lighten_u; - imp->combine_32[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_u; - imp->combine_32[PIXMAN_OP_COLOR_BURN] = combine_color_burn_u; imp->combine_32[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_u; - imp->combine_32[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_u; imp->combine_32[PIXMAN_OP_DIFFERENCE] = combine_difference_u; imp->combine_32[PIXMAN_OP_EXCLUSION] = combine_exclusion_u; - imp->combine_32[PIXMAN_OP_HSL_HUE] = combine_hsl_hue_u; - imp->combine_32[PIXMAN_OP_HSL_SATURATION] = combine_hsl_saturation_u; - imp->combine_32[PIXMAN_OP_HSL_COLOR] = combine_hsl_color_u; - imp->combine_32[PIXMAN_OP_HSL_LUMINOSITY] = combine_hsl_luminosity_u; /* Component alpha combiners */ imp->combine_32_ca[PIXMAN_OP_CLEAR] = combine_clear_ca; @@ -2531,51 +1177,13 @@ _pixman_setup_combiner_functions_32 (pixman_implementation_t *imp) imp->combine_32_ca[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_ca; imp->combine_32_ca[PIXMAN_OP_XOR] = combine_xor_ca; imp->combine_32_ca[PIXMAN_OP_ADD] = combine_add_ca; - imp->combine_32_ca[PIXMAN_OP_SATURATE] = combine_saturate_ca; - - /* Disjoint CA */ - imp->combine_32_ca[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_SRC] = combine_src_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_DST] = combine_dst; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_ca; - - /* Conjoint CA */ - imp->combine_32_ca[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_SRC] = combine_src_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_DST] = combine_dst; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_ca; - imp->combine_32_ca[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_ca; imp->combine_32_ca[PIXMAN_OP_MULTIPLY] = combine_multiply_ca; imp->combine_32_ca[PIXMAN_OP_SCREEN] = combine_screen_ca; imp->combine_32_ca[PIXMAN_OP_OVERLAY] = combine_overlay_ca; imp->combine_32_ca[PIXMAN_OP_DARKEN] = combine_darken_ca; imp->combine_32_ca[PIXMAN_OP_LIGHTEN] = combine_lighten_ca; - imp->combine_32_ca[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_ca; - imp->combine_32_ca[PIXMAN_OP_COLOR_BURN] = combine_color_burn_ca; imp->combine_32_ca[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_ca; - imp->combine_32_ca[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_ca; imp->combine_32_ca[PIXMAN_OP_DIFFERENCE] = combine_difference_ca; imp->combine_32_ca[PIXMAN_OP_EXCLUSION] = combine_exclusion_ca; - - /* It is not clear that these make sense, so make them noops for now */ - imp->combine_32_ca[PIXMAN_OP_HSL_HUE] = combine_dst; - imp->combine_32_ca[PIXMAN_OP_HSL_SATURATION] = combine_dst; - imp->combine_32_ca[PIXMAN_OP_HSL_COLOR] = combine_dst; - imp->combine_32_ca[PIXMAN_OP_HSL_LUMINOSITY] = combine_dst; } diff --git a/lib/pixman/pixman/pixman-fast-path.c b/lib/pixman/pixman/pixman-fast-path.c index c6e43de10..53d4a1f90 100644 --- a/lib/pixman/pixman/pixman-fast-path.c +++ b/lib/pixman/pixman/pixman-fast-path.c @@ -2343,6 +2343,8 @@ fast_fetch_bilinear_cover (pixman_iter_t *iter, const uint32_t *mask) int32_t dist_y; int i; + COMPILE_TIME_ASSERT (BILINEAR_INTERPOLATION_BITS < 8); + fx = info->x; ux = iter->image->common.transform->matrix[0][0]; diff --git a/lib/pixman/pixman/pixman-general.c b/lib/pixman/pixman/pixman-general.c index f82ea7d71..6141cb0a3 100644 --- a/lib/pixman/pixman/pixman-general.c +++ b/lib/pixman/pixman/pixman-general.c @@ -109,6 +109,20 @@ static const op_info_t op_flags[PIXMAN_N_OPERATORS] = #define SCANLINE_BUFFER_LENGTH 8192 +static pixman_bool_t +operator_needs_division (pixman_op_t op) +{ + static const uint8_t needs_division[] = + { + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, /* SATURATE */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, /* DISJOINT */ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, /* CONJOINT */ + 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 0, /* blend ops */ + }; + + return needs_division[op]; +} + static void general_composite_rect (pixman_implementation_t *imp, pixman_composite_info_t *info) @@ -124,9 +138,10 @@ general_composite_rect (pixman_implementation_t *imp, int Bpp; int i; - if ((src_image->common.flags & FAST_PATH_NARROW_FORMAT) && - (!mask_image || mask_image->common.flags & FAST_PATH_NARROW_FORMAT) && - (dest_image->common.flags & FAST_PATH_NARROW_FORMAT)) + if ((src_image->common.flags & FAST_PATH_NARROW_FORMAT) && + (!mask_image || mask_image->common.flags & FAST_PATH_NARROW_FORMAT) && + (dest_image->common.flags & FAST_PATH_NARROW_FORMAT) && + !(operator_needs_division (op))) { width_flag = ITER_NARROW; Bpp = 4; @@ -143,9 +158,9 @@ general_composite_rect (pixman_implementation_t *imp, if (width <= 0 || _pixman_multiply_overflows_int (width, Bpp * 3)) return; - if (width * Bpp * 3 > sizeof (stack_scanline_buffer) - 32 * 3) + if (width * Bpp * 3 > sizeof (stack_scanline_buffer) - 15 * 3) { - scanline_buffer = pixman_malloc_ab_plus_c (width, Bpp * 3, 32 * 3); + scanline_buffer = pixman_malloc_ab_plus_c (width, Bpp * 3, 15 * 3); if (!scanline_buffer) return; @@ -181,11 +196,7 @@ general_composite_rect (pixman_implementation_t *imp, mask_image = NULL; } - component_alpha = - mask_image && - mask_image->common.type == BITS && - mask_image->common.component_alpha && - PIXMAN_FORMAT_RGB (mask_image->bits.format); + component_alpha = mask_image && mask_image->common.component_alpha; _pixman_implementation_iter_init ( imp->toplevel, &mask_iter, diff --git a/lib/pixman/pixman/pixman-gradient-walker.c b/lib/pixman/pixman/pixman-gradient-walker.c index 5944a559a..822f8e62b 100644 --- a/lib/pixman/pixman/pixman-gradient-walker.c +++ b/lib/pixman/pixman/pixman-gradient-walker.c @@ -54,7 +54,7 @@ static void gradient_walker_reset (pixman_gradient_walker_t *walker, pixman_fixed_48_16_t pos) { - int32_t x, left_x, right_x; + int64_t x, left_x, right_x; pixman_color_t *left_c, *right_c; int n, count = walker->num_stops; pixman_gradient_stop_t *stops = walker->stops; diff --git a/lib/pixman/pixman/pixman-implementation.c b/lib/pixman/pixman/pixman-implementation.c index 588405451..2c7de4c68 100644 --- a/lib/pixman/pixman/pixman-implementation.c +++ b/lib/pixman/pixman/pixman-implementation.c @@ -380,6 +380,11 @@ _pixman_disabled (const char *name) return FALSE; } +static const pixman_fast_path_t empty_fast_path[] = +{ + { PIXMAN_OP_NONE } +}; + pixman_implementation_t * _pixman_choose_implementation (void) { @@ -397,5 +402,16 @@ _pixman_choose_implementation (void) imp = _pixman_implementation_create_noop (imp); + if (_pixman_disabled ("wholeops")) + { + pixman_implementation_t *cur; + + /* Disable all whole-operation paths except the general one, + * so that optimized iterators are used as much as possible. + */ + for (cur = imp; cur->fallback; cur = cur->fallback) + cur->fast_paths = empty_fast_path; + } + return imp; } diff --git a/lib/pixman/pixman/pixman-inlines.h b/lib/pixman/pixman/pixman-inlines.h index dd1c2f17f..1c8441d6d 100644 --- a/lib/pixman/pixman/pixman-inlines.h +++ b/lib/pixman/pixman/pixman-inlines.h @@ -747,7 +747,8 @@ fast_composite_scaled_nearest ## scale_func_name (pixman_implementation_t *imp, #define SIMPLE_NEAREST_SOLID_MASK_FAST_PATH(op,s,d,func) \ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_COVER (op,s,d,func), \ SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NONE (op,s,d,func), \ - SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_PAD (op,s,d,func) + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_PAD (op,s,d,func), \ + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NORMAL (op,s,d,func) /*****************************************************************************/ diff --git a/lib/pixman/pixman/pixman-mips-dspr2-asm.S b/lib/pixman/pixman/pixman-mips-dspr2-asm.S index 866e93e58..9dad163b7 100644 --- a/lib/pixman/pixman/pixman-mips-dspr2-asm.S +++ b/lib/pixman/pixman/pixman-mips-dspr2-asm.S @@ -26,7 +26,7 @@ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * - * Author: Nemanja Lukic (nlukic@mips.com) + * Author: Nemanja Lukic (nemanja.lukic@rt-rk.com) */ #include "pixman-private.h" diff --git a/lib/pixman/pixman/pixman-mips-dspr2-asm.h b/lib/pixman/pixman/pixman-mips-dspr2-asm.h index 11849bd66..e23856619 100644 --- a/lib/pixman/pixman/pixman-mips-dspr2-asm.h +++ b/lib/pixman/pixman/pixman-mips-dspr2-asm.h @@ -26,7 +26,7 @@ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * - * Author: Nemanja Lukic (nlukic@mips.com) + * Author: Nemanja Lukic (nemanja.lukic@rt-rk.com) */ #ifndef PIXMAN_MIPS_DSPR2_ASM_H @@ -72,10 +72,8 @@ #define LEAF_MIPS32R2(symbol) \ .globl symbol; \ .align 2; \ -#ifdef __ELF__ .hidden symbol; \ .type symbol, @function; \ -#endif .ent symbol, 0; \ symbol: .frame sp, 0, ra; \ .set push; \ diff --git a/lib/pixman/pixman/pixman-mips-dspr2.c b/lib/pixman/pixman/pixman-mips-dspr2.c index e10c9df0a..87969ae70 100644 --- a/lib/pixman/pixman/pixman-mips-dspr2.c +++ b/lib/pixman/pixman/pixman-mips-dspr2.c @@ -26,7 +26,7 @@ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * - * Author: Nemanja Lukic (nlukic@mips.com) + * Author: Nemanja Lukic (nemanja.lukic@rt-rk.com) */ #ifdef HAVE_CONFIG_H @@ -388,11 +388,11 @@ static const pixman_fast_path_t mips_dspr2_fast_paths[] = SIMPLE_NEAREST_FAST_PATH_PAD (SRC, r5g6b5, a8r8g8b8, mips_0565_8888), SIMPLE_NEAREST_FAST_PATH_PAD (SRC, b5g6r5, a8b8g8r8, mips_0565_8888), - PIXMAN_MIPS_SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, a8r8g8b8, r5g6b5, mips_8888_8_0565), - PIXMAN_MIPS_SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, a8b8g8r8, b5g6r5, mips_8888_8_0565), + SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, a8r8g8b8, r5g6b5, mips_8888_8_0565), + SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, a8b8g8r8, b5g6r5, mips_8888_8_0565), - PIXMAN_MIPS_SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, r5g6b5, r5g6b5, mips_0565_8_0565), - PIXMAN_MIPS_SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, b5g6r5, b5g6r5, mips_0565_8_0565), + SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, r5g6b5, r5g6b5, mips_0565_8_0565), + SIMPLE_NEAREST_A8_MASK_FAST_PATH (OVER, b5g6r5, b5g6r5, mips_0565_8_0565), SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, mips_8888_8888), SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, mips_8888_8888), diff --git a/lib/pixman/pixman/pixman-mips-dspr2.h b/lib/pixman/pixman/pixman-mips-dspr2.h index 955ed70b8..57b38359e 100644 --- a/lib/pixman/pixman/pixman-mips-dspr2.h +++ b/lib/pixman/pixman/pixman-mips-dspr2.h @@ -26,7 +26,7 @@ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * - * Author: Nemanja Lukic (nlukic@mips.com) + * Author: Nemanja Lukic (nemanja.lukic@rt-rk.com) */ #ifndef PIXMAN_MIPS_DSPR2_H @@ -328,12 +328,6 @@ FAST_NEAREST_MAINLOOP_COMMON (mips_##name##_pad_##op, \ scaled_nearest_scanline_mips_##name##_##op, \ src_type, uint8_t, dst_type, PAD, TRUE, FALSE) -/* Provide entries for the fast path table */ -#define PIXMAN_MIPS_SIMPLE_NEAREST_A8_MASK_FAST_PATH(op,s,d,func) \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_COVER (op,s,d,func), \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_NONE (op,s,d,func), \ - SIMPLE_NEAREST_A8_MASK_FAST_PATH_PAD (op,s,d,func) - /****************************************************************************/ #define PIXMAN_MIPS_BIND_SCALED_BILINEAR_SRC_DST(flags, name, op, \ diff --git a/lib/pixman/pixman/pixman-mmx.c b/lib/pixman/pixman/pixman-mmx.c index f9a92ce09..dec397432 100644 --- a/lib/pixman/pixman/pixman-mmx.c +++ b/lib/pixman/pixman/pixman-mmx.c @@ -89,21 +89,7 @@ _mm_mulhi_pu16 (__m64 __A, __m64 __B) return __A; } -# ifdef __OPTIMIZE__ -extern __inline __m64 __attribute__((__gnu_inline__, __always_inline__, __artificial__)) -_mm_shuffle_pi16 (__m64 __A, int8_t const __N) -{ - __m64 ret; - - asm ("pshufw %2, %1, %0\n\t" - : "=y" (ret) - : "y" (__A), "K" (__N) - ); - - return ret; -} -# else -# define _mm_shuffle_pi16(A, N) \ +# define _mm_shuffle_pi16(A, N) \ ({ \ __m64 ret; \ \ @@ -114,7 +100,6 @@ _mm_shuffle_pi16 (__m64 __A, int8_t const __N) \ ret; \ }) -# endif # endif #endif @@ -3555,6 +3540,105 @@ mmx_composite_over_reverse_n_8888 (pixman_implementation_t *imp, _mm_empty (); } +static force_inline void +scaled_nearest_scanline_mmx_8888_8888_OVER (uint32_t* pd, + const uint32_t* ps, + int32_t w, + pixman_fixed_t vx, + pixman_fixed_t unit_x, + pixman_fixed_t src_width_fixed, + pixman_bool_t fully_transparent_src) +{ + if (fully_transparent_src) + return; + + while (w) + { + __m64 d = load (pd); + __m64 s = load (ps + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + + store8888 (pd, core_combine_over_u_pixel_mmx (s, d)); + pd++; + + w--; + } + + _mm_empty (); +} + +FAST_NEAREST_MAINLOOP (mmx_8888_8888_cover_OVER, + scaled_nearest_scanline_mmx_8888_8888_OVER, + uint32_t, uint32_t, COVER) +FAST_NEAREST_MAINLOOP (mmx_8888_8888_none_OVER, + scaled_nearest_scanline_mmx_8888_8888_OVER, + uint32_t, uint32_t, NONE) +FAST_NEAREST_MAINLOOP (mmx_8888_8888_pad_OVER, + scaled_nearest_scanline_mmx_8888_8888_OVER, + uint32_t, uint32_t, PAD) +FAST_NEAREST_MAINLOOP (mmx_8888_8888_normal_OVER, + scaled_nearest_scanline_mmx_8888_8888_OVER, + uint32_t, uint32_t, NORMAL) + +static force_inline void +scaled_nearest_scanline_mmx_8888_n_8888_OVER (const uint32_t * mask, + uint32_t * dst, + const uint32_t * src, + int32_t w, + pixman_fixed_t vx, + pixman_fixed_t unit_x, + pixman_fixed_t src_width_fixed, + pixman_bool_t zero_src) +{ + __m64 mm_mask; + + if (zero_src || (*mask >> 24) == 0) + { + /* A workaround for https://gcc.gnu.org/PR47759 */ + _mm_empty (); + return; + } + + mm_mask = expand_alpha (load8888 (mask)); + + while (w) + { + uint32_t s = *(src + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + + if (s) + { + __m64 ms = load8888 (&s); + __m64 alpha = expand_alpha (ms); + __m64 dest = load8888 (dst); + + store8888 (dst, (in_over (ms, alpha, mm_mask, dest))); + } + + dst++; + w--; + } + + _mm_empty (); +} + +FAST_NEAREST_MAINLOOP_COMMON (mmx_8888_n_8888_cover_OVER, + scaled_nearest_scanline_mmx_8888_n_8888_OVER, + uint32_t, uint32_t, uint32_t, COVER, TRUE, TRUE) +FAST_NEAREST_MAINLOOP_COMMON (mmx_8888_n_8888_pad_OVER, + scaled_nearest_scanline_mmx_8888_n_8888_OVER, + uint32_t, uint32_t, uint32_t, PAD, TRUE, TRUE) +FAST_NEAREST_MAINLOOP_COMMON (mmx_8888_n_8888_none_OVER, + scaled_nearest_scanline_mmx_8888_n_8888_OVER, + uint32_t, uint32_t, uint32_t, NONE, TRUE, TRUE) +FAST_NEAREST_MAINLOOP_COMMON (mmx_8888_n_8888_normal_OVER, + scaled_nearest_scanline_mmx_8888_n_8888_OVER, + uint32_t, uint32_t, uint32_t, NORMAL, TRUE, TRUE) + #define BSHIFT ((1 << BILINEAR_INTERPOLATION_BITS)) #define BMSK (BSHIFT - 1) @@ -3995,6 +4079,16 @@ static const pixman_fast_path_t mmx_fast_paths[] = PIXMAN_STD_FAST_PATH (IN, a8, null, a8, mmx_composite_in_8_8 ), PIXMAN_STD_FAST_PATH (IN, solid, a8, a8, mmx_composite_in_n_8_8 ), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, mmx_8888_8888 ), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, mmx_8888_8888 ), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, mmx_8888_8888 ), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, mmx_8888_8888 ), + + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, mmx_8888_n_8888 ), + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, mmx_8888_n_8888 ), + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, mmx_8888_n_8888 ), + SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, mmx_8888_n_8888 ), + SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, mmx_8888_8888 ), SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, mmx_8888_8888 ), SIMPLE_BILINEAR_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, mmx_8888_8888 ), diff --git a/lib/pixman/pixman/pixman-private.h b/lib/pixman/pixman/pixman-private.h index 6ca13b216..73108a01d 100644 --- a/lib/pixman/pixman/pixman-private.h +++ b/lib/pixman/pixman/pixman-private.h @@ -7,7 +7,7 @@ * The defines which are shared between C and assembly code */ -/* bilinear interpolation precision (must be <= 8) */ +/* bilinear interpolation precision (must be < 8) */ #define BILINEAR_INTERPOLATION_BITS 7 #define BILINEAR_INTERPOLATION_RANGE (1 << BILINEAR_INTERPOLATION_BITS) @@ -345,8 +345,8 @@ typedef struct float r_s, r_b; float g_s, g_b; float b_s, b_b; - pixman_fixed_t left_x; - pixman_fixed_t right_x; + pixman_fixed_48_16_t left_x; + pixman_fixed_48_16_t right_x; pixman_gradient_stop_t *stops; int num_stops; diff --git a/lib/pixman/pixman/pixman-sse2.c b/lib/pixman/pixman/pixman-sse2.c index a6e780815..895510372 100644 --- a/lib/pixman/pixman/pixman-sse2.c +++ b/lib/pixman/pixman/pixman-sse2.c @@ -6274,31 +6274,15 @@ static const pixman_fast_path_t sse2_fast_paths[] = PIXMAN_STD_FAST_PATH (IN, solid, a8, a8, sse2_composite_in_n_8_8), PIXMAN_STD_FAST_PATH (IN, solid, null, a8, sse2_composite_in_n_8), - SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_COVER (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NONE (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_PAD (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NORMAL (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NORMAL (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NORMAL (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888), - SIMPLE_NEAREST_FAST_PATH_NORMAL (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_8888), SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_n_8888), SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_n_8888), SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_n_8888), SIMPLE_NEAREST_SOLID_MASK_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_n_8888), - SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NORMAL (OVER, a8r8g8b8, a8r8g8b8, sse2_8888_n_8888), - SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NORMAL (OVER, a8b8g8r8, a8b8g8r8, sse2_8888_n_8888), - SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NORMAL (OVER, a8r8g8b8, x8r8g8b8, sse2_8888_n_8888), - SIMPLE_NEAREST_SOLID_MASK_FAST_PATH_NORMAL (OVER, a8b8g8r8, x8b8g8r8, sse2_8888_n_8888), SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, sse2_8888_8888), SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, sse2_8888_8888), diff --git a/lib/pixman/pixman/pixman-vmx.c b/lib/pixman/pixman/pixman-vmx.c index c33631c0e..41efdcfa1 100644 --- a/lib/pixman/pixman/pixman-vmx.c +++ b/lib/pixman/pixman/pixman-vmx.c @@ -30,17 +30,41 @@ #endif #include "pixman-private.h" #include "pixman-combine32.h" +#include "pixman-inlines.h" #include #define AVV(x...) {x} +static vector unsigned int mask_ff000000; +static vector unsigned int mask_red; +static vector unsigned int mask_green; +static vector unsigned int mask_blue; +static vector unsigned int mask_565_fix_rb; +static vector unsigned int mask_565_fix_g; + static force_inline vector unsigned int splat_alpha (vector unsigned int pix) { +#ifdef WORDS_BIGENDIAN return vec_perm (pix, pix, (vector unsigned char)AVV ( 0x00, 0x00, 0x00, 0x00, 0x04, 0x04, 0x04, 0x04, 0x08, 0x08, 0x08, 0x08, 0x0C, 0x0C, 0x0C, 0x0C)); +#else + return vec_perm (pix, pix, + (vector unsigned char)AVV ( + 0x03, 0x03, 0x03, 0x03, 0x07, 0x07, 0x07, 0x07, + 0x0B, 0x0B, 0x0B, 0x0B, 0x0F, 0x0F, 0x0F, 0x0F)); +#endif +} + +static force_inline vector unsigned int +splat_pixel (vector unsigned int pix) +{ + return vec_perm (pix, pix, + (vector unsigned char)AVV ( + 0x00, 0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01, + 0x02, 0x02, 0x02, 0x02, 0x03, 0x03, 0x03, 0x03)); } static force_inline vector unsigned int @@ -50,12 +74,22 @@ pix_multiply (vector unsigned int p, vector unsigned int a) /* unpack to short */ hi = (vector unsigned short) +#ifdef WORDS_BIGENDIAN vec_mergeh ((vector unsigned char)AVV (0), (vector unsigned char)p); +#else + vec_mergeh ((vector unsigned char) p, + (vector unsigned char) AVV (0)); +#endif mod = (vector unsigned short) +#ifdef WORDS_BIGENDIAN vec_mergeh ((vector unsigned char)AVV (0), (vector unsigned char)a); +#else + vec_mergeh ((vector unsigned char) a, + (vector unsigned char) AVV (0)); +#endif hi = vec_mladd (hi, mod, (vector unsigned short) AVV (0x0080, 0x0080, 0x0080, 0x0080, @@ -67,11 +101,22 @@ pix_multiply (vector unsigned int p, vector unsigned int a) /* unpack to short */ lo = (vector unsigned short) +#ifdef WORDS_BIGENDIAN vec_mergel ((vector unsigned char)AVV (0), (vector unsigned char)p); +#else + vec_mergel ((vector unsigned char) p, + (vector unsigned char) AVV (0)); +#endif + mod = (vector unsigned short) +#ifdef WORDS_BIGENDIAN vec_mergel ((vector unsigned char)AVV (0), (vector unsigned char)a); +#else + vec_mergel ((vector unsigned char) a, + (vector unsigned char) AVV (0)); +#endif lo = vec_mladd (lo, mod, (vector unsigned short) AVV (0x0080, 0x0080, 0x0080, 0x0080, @@ -129,6 +174,7 @@ over (vector unsigned int src, over (pix_multiply (src, mask), \ pix_multiply (srca, mask), dest) +#ifdef WORDS_BIGENDIAN #define COMPUTE_SHIFT_MASK(source) \ source ## _mask = vec_lvsl (0, source); @@ -140,36 +186,305 @@ over (vector unsigned int src, mask ## _mask = vec_lvsl (0, mask); \ source ## _mask = vec_lvsl (0, source); -/* notice you have to declare temp vars... - * Note: tmp3 and tmp4 must remain untouched! - */ - -#define LOAD_VECTORS(dest, source) \ +#define LOAD_VECTOR(source) \ +do \ +{ \ + vector unsigned char tmp1, tmp2; \ tmp1 = (typeof(tmp1))vec_ld (0, source); \ tmp2 = (typeof(tmp2))vec_ld (15, source); \ - v ## source = (typeof(v ## source)) \ + v ## source = (typeof(v ## source)) \ vec_perm (tmp1, tmp2, source ## _mask); \ - v ## dest = (typeof(v ## dest))vec_ld (0, dest); +} while (0) -#define LOAD_VECTORSC(dest, source, mask) \ - tmp1 = (typeof(tmp1))vec_ld (0, source); \ - tmp2 = (typeof(tmp2))vec_ld (15, source); \ - v ## source = (typeof(v ## source)) \ - vec_perm (tmp1, tmp2, source ## _mask); \ - tmp1 = (typeof(tmp1))vec_ld (0, mask); \ +#define LOAD_VECTORS(dest, source) \ +do \ +{ \ + LOAD_VECTOR(source); \ v ## dest = (typeof(v ## dest))vec_ld (0, dest); \ - tmp2 = (typeof(tmp2))vec_ld (15, mask); \ - v ## mask = (typeof(v ## mask)) \ - vec_perm (tmp1, tmp2, mask ## _mask); +} while (0) + +#define LOAD_VECTORSC(dest, source, mask) \ +do \ +{ \ + LOAD_VECTORS(dest, source); \ + LOAD_VECTOR(mask); \ +} while (0) + +#define DECLARE_SRC_MASK_VAR vector unsigned char src_mask +#define DECLARE_MASK_MASK_VAR vector unsigned char mask_mask + +#else + +/* Now the COMPUTE_SHIFT_{MASK, MASKS, MASKC} below are just no-op. + * They are defined that way because little endian altivec can do unaligned + * reads natively and have no need for constructing the permutation pattern + * variables. + */ +#define COMPUTE_SHIFT_MASK(source) + +#define COMPUTE_SHIFT_MASKS(dest, source) + +#define COMPUTE_SHIFT_MASKC(dest, source, mask) + +# define LOAD_VECTOR(source) \ + v ## source = *((typeof(v ## source)*)source); + +# define LOAD_VECTORS(dest, source) \ + LOAD_VECTOR(source); \ + LOAD_VECTOR(dest); \ + +# define LOAD_VECTORSC(dest, source, mask) \ + LOAD_VECTORS(dest, source); \ + LOAD_VECTOR(mask); \ + +#define DECLARE_SRC_MASK_VAR +#define DECLARE_MASK_MASK_VAR + +#endif /* WORDS_BIGENDIAN */ #define LOAD_VECTORSM(dest, source, mask) \ - LOAD_VECTORSC (dest, source, mask) \ + LOAD_VECTORSC (dest, source, mask); \ v ## source = pix_multiply (v ## source, \ splat_alpha (v ## mask)); #define STORE_VECTOR(dest) \ vec_st ((vector unsigned int) v ## dest, 0, dest); +/* load 4 pixels from a 16-byte boundary aligned address */ +static force_inline vector unsigned int +load_128_aligned (const uint32_t* src) +{ + return *((vector unsigned int *) src); +} + +/* load 4 pixels from a unaligned address */ +static force_inline vector unsigned int +load_128_unaligned (const uint32_t* src) +{ + vector unsigned int vsrc; + DECLARE_SRC_MASK_VAR; + + COMPUTE_SHIFT_MASK (src); + LOAD_VECTOR (src); + + return vsrc; +} + +/* save 4 pixels on a 16-byte boundary aligned address */ +static force_inline void +save_128_aligned (uint32_t* data, + vector unsigned int vdata) +{ + STORE_VECTOR(data) +} + +static force_inline vector unsigned int +create_mask_1x32_128 (const uint32_t *src) +{ + vector unsigned int vsrc; + DECLARE_SRC_MASK_VAR; + + COMPUTE_SHIFT_MASK (src); + LOAD_VECTOR (src); + return vec_splat(vsrc, 0); +} + +static force_inline vector unsigned int +create_mask_32_128 (uint32_t mask) +{ + return create_mask_1x32_128(&mask); +} + +static force_inline vector unsigned int +unpacklo_128_16x8 (vector unsigned int data1, vector unsigned int data2) +{ + vector unsigned char lo; + + /* unpack to short */ + lo = (vector unsigned char) +#ifdef WORDS_BIGENDIAN + vec_mergel ((vector unsigned char) data2, + (vector unsigned char) data1); +#else + vec_mergel ((vector unsigned char) data1, + (vector unsigned char) data2); +#endif + + return (vector unsigned int) lo; +} + +static force_inline vector unsigned int +unpackhi_128_16x8 (vector unsigned int data1, vector unsigned int data2) +{ + vector unsigned char hi; + + /* unpack to short */ + hi = (vector unsigned char) +#ifdef WORDS_BIGENDIAN + vec_mergeh ((vector unsigned char) data2, + (vector unsigned char) data1); +#else + vec_mergeh ((vector unsigned char) data1, + (vector unsigned char) data2); +#endif + + return (vector unsigned int) hi; +} + +static force_inline vector unsigned int +unpacklo_128_8x16 (vector unsigned int data1, vector unsigned int data2) +{ + vector unsigned short lo; + + /* unpack to char */ + lo = (vector unsigned short) +#ifdef WORDS_BIGENDIAN + vec_mergel ((vector unsigned short) data2, + (vector unsigned short) data1); +#else + vec_mergel ((vector unsigned short) data1, + (vector unsigned short) data2); +#endif + + return (vector unsigned int) lo; +} + +static force_inline vector unsigned int +unpackhi_128_8x16 (vector unsigned int data1, vector unsigned int data2) +{ + vector unsigned short hi; + + /* unpack to char */ + hi = (vector unsigned short) +#ifdef WORDS_BIGENDIAN + vec_mergeh ((vector unsigned short) data2, + (vector unsigned short) data1); +#else + vec_mergeh ((vector unsigned short) data1, + (vector unsigned short) data2); +#endif + + return (vector unsigned int) hi; +} + +static force_inline void +unpack_128_2x128 (vector unsigned int data1, vector unsigned int data2, + vector unsigned int* data_lo, vector unsigned int* data_hi) +{ + *data_lo = unpacklo_128_16x8(data1, data2); + *data_hi = unpackhi_128_16x8(data1, data2); +} + +static force_inline void +unpack_128_2x128_16 (vector unsigned int data1, vector unsigned int data2, + vector unsigned int* data_lo, vector unsigned int* data_hi) +{ + *data_lo = unpacklo_128_8x16(data1, data2); + *data_hi = unpackhi_128_8x16(data1, data2); +} + +static force_inline vector unsigned int +unpack_565_to_8888 (vector unsigned int lo) +{ + vector unsigned int r, g, b, rb, t; + + r = vec_and (vec_sl(lo, create_mask_32_128(8)), mask_red); + g = vec_and (vec_sl(lo, create_mask_32_128(5)), mask_green); + b = vec_and (vec_sl(lo, create_mask_32_128(3)), mask_blue); + + rb = vec_or (r, b); + t = vec_and (rb, mask_565_fix_rb); + t = vec_sr (t, create_mask_32_128(5)); + rb = vec_or (rb, t); + + t = vec_and (g, mask_565_fix_g); + t = vec_sr (t, create_mask_32_128(6)); + g = vec_or (g, t); + + return vec_or (rb, g); +} + +static force_inline int +is_opaque (vector unsigned int x) +{ + uint32_t cmp_result; + vector bool int ffs = vec_cmpeq(x, x); + + cmp_result = vec_all_eq(x, ffs); + + return (cmp_result & 0x8888) == 0x8888; +} + +static force_inline int +is_zero (vector unsigned int x) +{ + uint32_t cmp_result; + + cmp_result = vec_all_eq(x, (vector unsigned int) AVV(0)); + + return cmp_result == 0xffff; +} + +static force_inline int +is_transparent (vector unsigned int x) +{ + uint32_t cmp_result; + + cmp_result = vec_all_eq(x, (vector unsigned int) AVV(0)); + return (cmp_result & 0x8888) == 0x8888; +} + +static force_inline uint32_t +core_combine_over_u_pixel_vmx (uint32_t src, uint32_t dst) +{ + uint32_t a; + + a = ALPHA_8(src); + + if (a == 0xff) + { + return src; + } + else if (src) + { + UN8x4_MUL_UN8_ADD_UN8x4(dst, (~a & MASK), src); + } + + return dst; +} + +static force_inline uint32_t +combine1 (const uint32_t *ps, const uint32_t *pm) +{ + uint32_t s = *ps; + + if (pm) + UN8x4_MUL_UN8(s, ALPHA_8(*pm)); + + return s; +} + +static force_inline vector unsigned int +combine4 (const uint32_t* ps, const uint32_t* pm) +{ + vector unsigned int src, msk; + + if (pm) + { + msk = load_128_unaligned(pm); + + if (is_transparent(msk)) + return (vector unsigned int) AVV(0); + } + + src = load_128_unaligned(ps); + + if (pm) + src = pix_multiply(src, msk); + + return src; +} + static void vmx_combine_over_u_no_mask (uint32_t * dest, const uint32_t *src, @@ -177,7 +492,7 @@ vmx_combine_over_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -227,7 +542,8 @@ vmx_combine_over_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -298,7 +614,7 @@ vmx_combine_over_reverse_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -346,7 +662,8 @@ vmx_combine_over_reverse_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -414,7 +731,7 @@ vmx_combine_in_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -459,7 +776,8 @@ vmx_combine_in_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -524,7 +842,7 @@ vmx_combine_in_reverse_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -571,7 +889,8 @@ vmx_combine_in_reverse_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -638,7 +957,7 @@ vmx_combine_out_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -685,7 +1004,8 @@ vmx_combine_out_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -750,7 +1070,7 @@ vmx_combine_out_reverse_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -798,7 +1118,8 @@ vmx_combine_out_reverse_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -865,7 +1186,7 @@ vmx_combine_atop_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -917,7 +1238,8 @@ vmx_combine_atop_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -993,7 +1315,7 @@ vmx_combine_atop_reverse_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1045,7 +1367,8 @@ vmx_combine_atop_reverse_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1121,7 +1444,7 @@ vmx_combine_xor_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1173,7 +1496,8 @@ vmx_combine_xor_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1249,7 +1573,7 @@ vmx_combine_add_u_no_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc; - vector unsigned char tmp1, tmp2, src_mask; + DECLARE_SRC_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1295,7 +1619,8 @@ vmx_combine_add_u_mask (uint32_t * dest, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, src_mask, mask_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1363,7 +1688,8 @@ vmx_combine_src_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1413,7 +1739,8 @@ vmx_combine_over_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1471,7 +1798,8 @@ vmx_combine_over_reverse_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1527,7 +1855,8 @@ vmx_combine_in_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1581,7 +1910,8 @@ vmx_combine_in_reverse_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1636,7 +1966,8 @@ vmx_combine_out_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1693,7 +2024,8 @@ vmx_combine_out_reverse_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1750,7 +2082,8 @@ vmx_combine_atop_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask, vsrca; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1816,7 +2149,8 @@ vmx_combine_atop_reverse_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1879,7 +2213,8 @@ vmx_combine_xor_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1942,7 +2277,8 @@ vmx_combine_add_ca (pixman_implementation_t *imp, { int i; vector unsigned int vdest, vsrc, vmask; - vector unsigned char tmp1, tmp2, mask_mask, src_mask; + DECLARE_SRC_MASK_VAR; + DECLARE_MASK_MASK_VAR; while (width && ((uintptr_t)dest & 15)) { @@ -1986,16 +2322,809 @@ vmx_combine_add_ca (pixman_implementation_t *imp, } } +static void +vmx_composite_over_n_8_8888 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint32_t src, srca; + uint32_t *dst_line, *dst; + uint8_t *mask_line; + int dst_stride, mask_stride; + int32_t w; + uint32_t m, d, s, ia; + + vector unsigned int vsrc, valpha, vmask, vdst; + + src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format); + + srca = ALPHA_8(src); + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE ( + mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + vsrc = (vector unsigned int) {src, src, src, src}; + valpha = splat_alpha(vsrc); + + while (height--) + { + const uint8_t *pm = mask_line; + dst = dst_line; + dst_line += dst_stride; + mask_line += mask_stride; + w = width; + + while (w && (uintptr_t)dst & 15) + { + s = src; + m = *pm++; + + if (m) + { + d = *dst; + UN8x4_MUL_UN8 (s, m); + ia = ALPHA_8 (~s); + UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s); + *dst = d; + } + + w--; + dst++; + } + + while (w >= 4) + { + m = *((uint32_t*)pm); + + if (srca == 0xff && m == 0xffffffff) + { + save_128_aligned(dst, vsrc); + } + else if (m) + { + vmask = splat_pixel((vector unsigned int) {m, m, m, m}); + + /* dst is 16-byte aligned */ + vdst = in_over (vsrc, valpha, vmask, load_128_aligned (dst)); + + save_128_aligned(dst, vdst); + } + + w -= 4; + dst += 4; + pm += 4; + } + + while (w) + { + s = src; + m = *pm++; + + if (m) + { + d = *dst; + UN8x4_MUL_UN8 (s, m); + ia = ALPHA_8 (~s); + UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s); + *dst = d; + } + + w--; + dst++; + } + } + +} + +static pixman_bool_t +vmx_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t filler) +{ + uint32_t byte_width; + uint8_t *byte_line; + + vector unsigned int vfiller; + + if (bpp == 8) + { + uint8_t b; + uint16_t w; + + stride = stride * (int) sizeof (uint32_t) / 1; + byte_line = (uint8_t *)(((uint8_t *)bits) + stride * y + x); + byte_width = width; + stride *= 1; + + b = filler & 0xff; + w = (b << 8) | b; + filler = (w << 16) | w; + } + else if (bpp == 16) + { + stride = stride * (int) sizeof (uint32_t) / 2; + byte_line = (uint8_t *)(((uint16_t *)bits) + stride * y + x); + byte_width = 2 * width; + stride *= 2; + + filler = (filler & 0xffff) * 0x00010001; + } + else if (bpp == 32) + { + stride = stride * (int) sizeof (uint32_t) / 4; + byte_line = (uint8_t *)(((uint32_t *)bits) + stride * y + x); + byte_width = 4 * width; + stride *= 4; + } + else + { + return FALSE; + } + + vfiller = create_mask_1x32_128(&filler); + + while (height--) + { + int w; + uint8_t *d = byte_line; + byte_line += stride; + w = byte_width; + + if (w >= 1 && ((uintptr_t)d & 1)) + { + *(uint8_t *)d = filler; + w -= 1; + d += 1; + } + + while (w >= 2 && ((uintptr_t)d & 3)) + { + *(uint16_t *)d = filler; + w -= 2; + d += 2; + } + + while (w >= 4 && ((uintptr_t)d & 15)) + { + *(uint32_t *)d = filler; + + w -= 4; + d += 4; + } + + while (w >= 128) + { + vec_st(vfiller, 0, (uint32_t *) d); + vec_st(vfiller, 0, (uint32_t *) d + 4); + vec_st(vfiller, 0, (uint32_t *) d + 8); + vec_st(vfiller, 0, (uint32_t *) d + 12); + vec_st(vfiller, 0, (uint32_t *) d + 16); + vec_st(vfiller, 0, (uint32_t *) d + 20); + vec_st(vfiller, 0, (uint32_t *) d + 24); + vec_st(vfiller, 0, (uint32_t *) d + 28); + + d += 128; + w -= 128; + } + + if (w >= 64) + { + vec_st(vfiller, 0, (uint32_t *) d); + vec_st(vfiller, 0, (uint32_t *) d + 4); + vec_st(vfiller, 0, (uint32_t *) d + 8); + vec_st(vfiller, 0, (uint32_t *) d + 12); + + d += 64; + w -= 64; + } + + if (w >= 32) + { + vec_st(vfiller, 0, (uint32_t *) d); + vec_st(vfiller, 0, (uint32_t *) d + 4); + + d += 32; + w -= 32; + } + + if (w >= 16) + { + vec_st(vfiller, 0, (uint32_t *) d); + + d += 16; + w -= 16; + } + + while (w >= 4) + { + *(uint32_t *)d = filler; + + w -= 4; + d += 4; + } + + if (w >= 2) + { + *(uint16_t *)d = filler; + w -= 2; + d += 2; + } + + if (w >= 1) + { + *(uint8_t *)d = filler; + w -= 1; + d += 1; + } + } + + return TRUE; +} + +static void +vmx_composite_src_x888_8888 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int32_t w; + int dst_stride, src_stride; + + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE ( + src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (uintptr_t)dst & 15) + { + *dst++ = *src++ | 0xff000000; + w--; + } + + while (w >= 16) + { + vector unsigned int vmx_src1, vmx_src2, vmx_src3, vmx_src4; + + vmx_src1 = load_128_unaligned (src); + vmx_src2 = load_128_unaligned (src + 4); + vmx_src3 = load_128_unaligned (src + 8); + vmx_src4 = load_128_unaligned (src + 12); + + save_128_aligned (dst, vec_or (vmx_src1, mask_ff000000)); + save_128_aligned (dst + 4, vec_or (vmx_src2, mask_ff000000)); + save_128_aligned (dst + 8, vec_or (vmx_src3, mask_ff000000)); + save_128_aligned (dst + 12, vec_or (vmx_src4, mask_ff000000)); + + dst += 16; + src += 16; + w -= 16; + } + + while (w) + { + *dst++ = *src++ | 0xff000000; + w--; + } + } +} + +static void +vmx_composite_over_n_8888 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint32_t *dst_line, *dst; + uint32_t src, ia; + int i, w, dst_stride; + vector unsigned int vdst, vsrc, via; + + src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format); + + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + + vsrc = (vector unsigned int){src, src, src, src}; + via = negate (splat_alpha (vsrc)); + ia = ALPHA_8 (~src); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + w = width; + + while (w && ((uintptr_t)dst & 15)) + { + uint32_t d = *dst; + UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, src); + *dst++ = d; + w--; + } + + for (i = w / 4; i > 0; i--) + { + vdst = pix_multiply (load_128_aligned (dst), via); + save_128_aligned (dst, pix_add (vsrc, vdst)); + dst += 4; + } + + for (i = w % 4; --i >= 0;) + { + uint32_t d = dst[i]; + UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, src); + dst[i] = d; + } + } +} + +static void +vmx_composite_over_8888_8888 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + int dst_stride, src_stride; + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE ( + src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + dst = dst_line; + src = src_line; + + while (height--) + { + vmx_combine_over_u (imp, op, dst, src, NULL, width); + + dst += dst_stride; + src += src_stride; + } +} + +static void +vmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint32_t src, ia; + uint32_t *dst_line, d; + uint32_t *mask_line, m; + uint32_t pack_cmp; + int dst_stride, mask_stride; + + vector unsigned int vsrc, valpha, vmask, vdest; + + src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format); + + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE ( + mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + vsrc = (vector unsigned int) {src, src, src, src}; + valpha = splat_alpha(vsrc); + ia = ALPHA_8 (src); + + while (height--) + { + int w = width; + const uint32_t *pm = (uint32_t *)mask_line; + uint32_t *pd = (uint32_t *)dst_line; + uint32_t s; + + dst_line += dst_stride; + mask_line += mask_stride; + + while (w && (uintptr_t)pd & 15) + { + s = src; + m = *pm++; + + if (m) + { + d = *pd; + UN8x4_MUL_UN8x4 (s, m); + UN8x4_MUL_UN8 (m, ia); + m = ~m; + UN8x4_MUL_UN8x4_ADD_UN8x4 (d, m, s); + *pd = d; + } + + pd++; + w--; + } + + while (w >= 4) + { + /* pm is NOT necessarily 16-byte aligned */ + vmask = load_128_unaligned (pm); + + pack_cmp = vec_all_eq(vmask, (vector unsigned int) AVV(0)); + + /* if all bits in mask are zero, pack_cmp is not 0 */ + if (pack_cmp == 0) + { + /* pd is 16-byte aligned */ + vdest = in_over (vsrc, valpha, vmask, load_128_aligned (pd)); + + save_128_aligned(pd, vdest); + } + + pd += 4; + pm += 4; + w -= 4; + } + + while (w) + { + s = src; + m = *pm++; + + if (m) + { + d = *pd; + UN8x4_MUL_UN8x4 (s, m); + UN8x4_MUL_UN8 (m, ia); + m = ~m; + UN8x4_MUL_UN8x4_ADD_UN8x4 (d, m, s); + *pd = d; + } + + pd++; + w--; + } + } +} + +static void +vmx_composite_add_8_8 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint8_t *dst_line, *dst; + uint8_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + uint16_t t; + + PIXMAN_IMAGE_GET_LINE ( + src_image, src_x, src_y, uint8_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + src = src_line; + + dst_line += dst_stride; + src_line += src_stride; + w = width; + + /* Small head */ + while (w && (uintptr_t)dst & 3) + { + t = (*dst) + (*src++); + *dst++ = t | (0 - (t >> 8)); + w--; + } + + vmx_combine_add_u (imp, op, + (uint32_t*)dst, (uint32_t*)src, NULL, w >> 2); + + /* Small tail */ + dst += w & 0xfffc; + src += w & 0xfffc; + + w &= 3; + + while (w) + { + t = (*dst) + (*src++); + *dst++ = t | (0 - (t >> 8)); + w--; + } + } +} + +static void +vmx_composite_add_8888_8888 (pixman_implementation_t *imp, + pixman_composite_info_t *info) +{ + PIXMAN_COMPOSITE_ARGS (info); + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + + PIXMAN_IMAGE_GET_LINE ( + src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE ( + dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + + vmx_combine_add_u (imp, op, dst, src, NULL, width); + } +} + +static force_inline void +scaled_nearest_scanline_vmx_8888_8888_OVER (uint32_t* pd, + const uint32_t* ps, + int32_t w, + pixman_fixed_t vx, + pixman_fixed_t unit_x, + pixman_fixed_t src_width_fixed, + pixman_bool_t fully_transparent_src) +{ + uint32_t s, d; + const uint32_t* pm = NULL; + + vector unsigned int vsrc, vdst; + + if (fully_transparent_src) + return; + + /* Align dst on a 16-byte boundary */ + while (w && ((uintptr_t)pd & 15)) + { + d = *pd; + s = combine1 (ps + pixman_fixed_to_int (vx), pm); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + + *pd++ = core_combine_over_u_pixel_vmx (s, d); + if (pm) + pm++; + w--; + } + + while (w >= 4) + { + vector unsigned int tmp; + uint32_t tmp1, tmp2, tmp3, tmp4; + + tmp1 = *(ps + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + tmp2 = *(ps + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + tmp3 = *(ps + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + tmp4 = *(ps + pixman_fixed_to_int (vx)); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + + tmp[0] = tmp1; + tmp[1] = tmp2; + tmp[2] = tmp3; + tmp[3] = tmp4; + + vsrc = combine4 ((const uint32_t *) &tmp, pm); + + if (is_opaque (vsrc)) + { + save_128_aligned (pd, vsrc); + } + else if (!is_zero (vsrc)) + { + vdst = over(vsrc, splat_alpha(vsrc), load_128_aligned (pd)); + + save_128_aligned (pd, vdst); + } + + w -= 4; + pd += 4; + if (pm) + pm += 4; + } + + while (w) + { + d = *pd; + s = combine1 (ps + pixman_fixed_to_int (vx), pm); + vx += unit_x; + while (vx >= 0) + vx -= src_width_fixed; + + *pd++ = core_combine_over_u_pixel_vmx (s, d); + if (pm) + pm++; + + w--; + } +} + +FAST_NEAREST_MAINLOOP (vmx_8888_8888_cover_OVER, + scaled_nearest_scanline_vmx_8888_8888_OVER, + uint32_t, uint32_t, COVER) +FAST_NEAREST_MAINLOOP (vmx_8888_8888_none_OVER, + scaled_nearest_scanline_vmx_8888_8888_OVER, + uint32_t, uint32_t, NONE) +FAST_NEAREST_MAINLOOP (vmx_8888_8888_pad_OVER, + scaled_nearest_scanline_vmx_8888_8888_OVER, + uint32_t, uint32_t, PAD) +FAST_NEAREST_MAINLOOP (vmx_8888_8888_normal_OVER, + scaled_nearest_scanline_vmx_8888_8888_OVER, + uint32_t, uint32_t, NORMAL) + static const pixman_fast_path_t vmx_fast_paths[] = { + PIXMAN_STD_FAST_PATH (OVER, solid, null, a8r8g8b8, vmx_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, null, x8r8g8b8, vmx_composite_over_n_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, a8r8g8b8, vmx_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, vmx_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, vmx_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, x8b8g8r8, vmx_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, vmx_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8r8g8b8, vmx_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, vmx_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, vmx_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, vmx_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, vmx_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, vmx_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, x8b8g8r8, vmx_composite_over_n_8888_8888_ca), + + /* PIXMAN_OP_ADD */ + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, vmx_composite_add_8_8), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, vmx_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, vmx_composite_add_8888_8888), + + /* PIXMAN_OP_SRC */ + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, a8r8g8b8, vmx_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, a8b8g8r8, vmx_composite_src_x888_8888), + + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, vmx_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, vmx_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, vmx_8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, vmx_8888_8888), + { PIXMAN_OP_NONE }, }; +static uint32_t * +vmx_fetch_x8r8g8b8 (pixman_iter_t *iter, const uint32_t *mask) +{ + int w = iter->width; + vector unsigned int ff000000 = mask_ff000000; + uint32_t *dst = iter->buffer; + uint32_t *src = (uint32_t *)iter->bits; + + iter->bits += iter->stride; + + while (w && ((uintptr_t)dst) & 0x0f) + { + *dst++ = (*src++) | 0xff000000; + w--; + } + + while (w >= 4) + { + save_128_aligned(dst, vec_or(load_128_unaligned(src), ff000000)); + + dst += 4; + src += 4; + w -= 4; + } + + while (w) + { + *dst++ = (*src++) | 0xff000000; + w--; + } + + return iter->buffer; +} + +static uint32_t * +vmx_fetch_a8 (pixman_iter_t *iter, const uint32_t *mask) +{ + int w = iter->width; + uint32_t *dst = iter->buffer; + uint8_t *src = iter->bits; + vector unsigned int vmx0, vmx1, vmx2, vmx3, vmx4, vmx5, vmx6; + + iter->bits += iter->stride; + + while (w && (((uintptr_t)dst) & 15)) + { + *dst++ = *(src++) << 24; + w--; + } + + while (w >= 16) + { + vmx0 = load_128_unaligned((uint32_t *) src); + + unpack_128_2x128((vector unsigned int) AVV(0), vmx0, &vmx1, &vmx2); + unpack_128_2x128_16((vector unsigned int) AVV(0), vmx1, &vmx3, &vmx4); + unpack_128_2x128_16((vector unsigned int) AVV(0), vmx2, &vmx5, &vmx6); + + save_128_aligned(dst, vmx6); + save_128_aligned((dst + 4), vmx5); + save_128_aligned((dst + 8), vmx4); + save_128_aligned((dst + 12), vmx3); + + dst += 16; + src += 16; + w -= 16; + } + + while (w) + { + *dst++ = *(src++) << 24; + w--; + } + + return iter->buffer; +} + +#define IMAGE_FLAGS \ + (FAST_PATH_STANDARD_FLAGS | FAST_PATH_ID_TRANSFORM | \ + FAST_PATH_BITS_IMAGE | FAST_PATH_SAMPLES_COVER_CLIP_NEAREST) + +static const pixman_iter_info_t vmx_iters[] = +{ + { PIXMAN_x8r8g8b8, IMAGE_FLAGS, ITER_NARROW, + _pixman_iter_init_bits_stride, vmx_fetch_x8r8g8b8, NULL + }, + { PIXMAN_a8, IMAGE_FLAGS, ITER_NARROW, + _pixman_iter_init_bits_stride, vmx_fetch_a8, NULL + }, + { PIXMAN_null }, +}; + pixman_implementation_t * _pixman_implementation_create_vmx (pixman_implementation_t *fallback) { pixman_implementation_t *imp = _pixman_implementation_create (fallback, vmx_fast_paths); + /* VMX constants */ + mask_ff000000 = create_mask_32_128 (0xff000000); + mask_red = create_mask_32_128 (0x00f80000); + mask_green = create_mask_32_128 (0x0000fc00); + mask_blue = create_mask_32_128 (0x000000f8); + mask_565_fix_rb = create_mask_32_128 (0x00e000e0); + mask_565_fix_g = create_mask_32_128 (0x0000c000); + /* Set up function pointers */ imp->combine_32[PIXMAN_OP_OVER] = vmx_combine_over_u; @@ -2022,5 +3151,9 @@ _pixman_implementation_create_vmx (pixman_implementation_t *fallback) imp->combine_32_ca[PIXMAN_OP_XOR] = vmx_combine_xor_ca; imp->combine_32_ca[PIXMAN_OP_ADD] = vmx_combine_add_ca; + imp->fill = vmx_fill; + + imp->iter_info = vmx_iters; + return imp; } diff --git a/lib/pixman/pixman/pixman.c b/lib/pixman/pixman/pixman.c index 9555ceaaf..f932eac3c 100644 --- a/lib/pixman/pixman/pixman.c +++ b/lib/pixman/pixman/pixman.c @@ -325,18 +325,20 @@ _pixman_compute_composite_region32 (pixman_region32_t * region, return TRUE; } -typedef struct +typedef struct box_48_16 box_48_16_t; + +struct box_48_16 { - pixman_fixed_48_16_t x1; - pixman_fixed_48_16_t y1; - pixman_fixed_48_16_t x2; - pixman_fixed_48_16_t y2; -} box_48_16_t; + pixman_fixed_48_16_t x1; + pixman_fixed_48_16_t y1; + pixman_fixed_48_16_t x2; + pixman_fixed_48_16_t y2; +}; static pixman_bool_t -compute_transformed_extents (pixman_transform_t *transform, +compute_transformed_extents (pixman_transform_t *transform, const pixman_box32_t *extents, - box_48_16_t *transformed) + box_48_16_t *transformed) { pixman_fixed_48_16_t tx1, ty1, tx2, ty2; pixman_fixed_t x1, y1, x2, y2; @@ -495,21 +497,12 @@ analyze_extent (pixman_image_t *image, if (!compute_transformed_extents (transform, extents, &transformed)) return FALSE; - /* Expand the source area by a tiny bit so account of different rounding that - * may happen during sampling. Note that (8 * pixman_fixed_e) is very far from - * 0.5 so this won't cause the area computed to be overly pessimistic. - */ - transformed.x1 -= 8 * pixman_fixed_e; - transformed.y1 -= 8 * pixman_fixed_e; - transformed.x2 += 8 * pixman_fixed_e; - transformed.y2 += 8 * pixman_fixed_e; - if (image->common.type == BITS) { - if (pixman_fixed_to_int (transformed.x1) >= 0 && - pixman_fixed_to_int (transformed.y1) >= 0 && - pixman_fixed_to_int (transformed.x2) < image->bits.width && - pixman_fixed_to_int (transformed.y2) < image->bits.height) + if (pixman_fixed_to_int (transformed.x1 - pixman_fixed_e) >= 0 && + pixman_fixed_to_int (transformed.y1 - pixman_fixed_e) >= 0 && + pixman_fixed_to_int (transformed.x2 - pixman_fixed_e) < image->bits.width && + pixman_fixed_to_int (transformed.y2 - pixman_fixed_e) < image->bits.height) { *flags |= FAST_PATH_SAMPLES_COVER_CLIP_NEAREST; } diff --git a/lib/pixman/test/Makefile.in b/lib/pixman/test/Makefile.in index d7fa0d297..4461a93b9 100644 --- a/lib/pixman/test/Makefile.in +++ b/lib/pixman/test/Makefile.in @@ -74,20 +74,25 @@ AM_V_lt = $(am__v_lt_@AM_V@) am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@) am__v_lt_0 = --silent am__v_lt_1 = -am__EXEEXT_1 = prng-test$(EXEEXT) a1-trap-test$(EXEEXT) \ - pdf-op-test$(EXEEXT) region-test$(EXEEXT) \ - region-translate-test$(EXEEXT) combiner-test$(EXEEXT) \ - pixel-test$(EXEEXT) fetch-test$(EXEEXT) rotate-test$(EXEEXT) \ - oob-test$(EXEEXT) infinite-loop$(EXEEXT) trap-crasher$(EXEEXT) \ - alpha-loop$(EXEEXT) thread-test$(EXEEXT) \ - scaling-crash-test$(EXEEXT) scaling-helpers-test$(EXEEXT) \ - gradient-crash-test$(EXEEXT) region-contains-test$(EXEEXT) \ - alphamap$(EXEEXT) matrix-test$(EXEEXT) stress-test$(EXEEXT) \ - composite-traps-test$(EXEEXT) blitters-test$(EXEEXT) \ - glyph-test$(EXEEXT) scaling-test$(EXEEXT) affine-test$(EXEEXT) \ - composite$(EXEEXT) +am__EXEEXT_1 = oob-test$(EXEEXT) infinite-loop$(EXEEXT) \ + trap-crasher$(EXEEXT) fence-image-self-test$(EXEEXT) \ + region-translate-test$(EXEEXT) fetch-test$(EXEEXT) \ + a1-trap-test$(EXEEXT) prng-test$(EXEEXT) \ + radial-invalid$(EXEEXT) pdf-op-test$(EXEEXT) \ + region-test$(EXEEXT) combiner-test$(EXEEXT) \ + scaling-crash-test$(EXEEXT) alpha-loop$(EXEEXT) \ + scaling-helpers-test$(EXEEXT) thread-test$(EXEEXT) \ + rotate-test$(EXEEXT) alphamap$(EXEEXT) \ + gradient-crash-test$(EXEEXT) pixel-test$(EXEEXT) \ + matrix-test$(EXEEXT) composite-traps-test$(EXEEXT) \ + region-contains-test$(EXEEXT) glyph-test$(EXEEXT) \ + solid-test$(EXEEXT) stress-test$(EXEEXT) cover-test$(EXEEXT) \ + blitters-test$(EXEEXT) affine-test$(EXEEXT) \ + scaling-test$(EXEEXT) composite$(EXEEXT) \ + tolerance-test$(EXEEXT) am__EXEEXT_2 = lowlevel-blt-bench$(EXEEXT) radial-perf-test$(EXEEXT) \ - check-formats$(EXEEXT) scaling-bench$(EXEEXT) + check-formats$(EXEEXT) scaling-bench$(EXEEXT) \ + affine-bench$(EXEEXT) PROGRAMS = $(noinst_PROGRAMS) a1_trap_test_SOURCES = a1-trap-test.c a1_trap_test_OBJECTS = a1-trap-test.$(OBJEXT) @@ -96,6 +101,12 @@ am__DEPENDENCIES_1 = a1_trap_test_DEPENDENCIES = libutils.la \ $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ $(am__DEPENDENCIES_1) +affine_bench_SOURCES = affine-bench.c +affine_bench_OBJECTS = affine-bench.$(OBJEXT) +affine_bench_LDADD = $(LDADD) +affine_bench_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) affine_test_SOURCES = affine-test.c affine_test_OBJECTS = affine-test.$(OBJEXT) affine_test_LDADD = $(LDADD) @@ -144,6 +155,18 @@ composite_traps_test_LDADD = $(LDADD) composite_traps_test_DEPENDENCIES = libutils.la \ $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ $(am__DEPENDENCIES_1) +cover_test_SOURCES = cover-test.c +cover_test_OBJECTS = cover-test.$(OBJEXT) +cover_test_LDADD = $(LDADD) +cover_test_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) +fence_image_self_test_SOURCES = fence-image-self-test.c +fence_image_self_test_OBJECTS = fence-image-self-test.$(OBJEXT) +fence_image_self_test_LDADD = $(LDADD) +fence_image_self_test_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) fetch_test_SOURCES = fetch-test.c fetch_test_OBJECTS = fetch-test.$(OBJEXT) fetch_test_LDADD = $(LDADD) @@ -204,6 +227,12 @@ prng_test_LDADD = $(LDADD) prng_test_DEPENDENCIES = libutils.la \ $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ $(am__DEPENDENCIES_1) +radial_invalid_SOURCES = radial-invalid.c +radial_invalid_OBJECTS = radial-invalid.$(OBJEXT) +radial_invalid_LDADD = $(LDADD) +radial_invalid_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) radial_perf_test_SOURCES = radial-perf-test.c radial_perf_test_OBJECTS = radial-perf-test.$(OBJEXT) radial_perf_test_LDADD = $(LDADD) @@ -258,6 +287,12 @@ scaling_test_LDADD = $(LDADD) scaling_test_DEPENDENCIES = libutils.la \ $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ $(am__DEPENDENCIES_1) +solid_test_SOURCES = solid-test.c +solid_test_OBJECTS = solid-test.$(OBJEXT) +solid_test_LDADD = $(LDADD) +solid_test_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) stress_test_SOURCES = stress-test.c stress_test_OBJECTS = stress-test.$(OBJEXT) stress_test_LDADD = $(LDADD) @@ -270,6 +305,12 @@ thread_test_LDADD = $(LDADD) thread_test_DEPENDENCIES = libutils.la \ $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ $(am__DEPENDENCIES_1) +tolerance_test_SOURCES = tolerance-test.c +tolerance_test_OBJECTS = tolerance-test.$(OBJEXT) +tolerance_test_LDADD = $(LDADD) +tolerance_test_DEPENDENCIES = libutils.la \ + $(top_builddir)/pixman/libpixman-1.la $(am__DEPENDENCIES_1) \ + $(am__DEPENDENCIES_1) trap_crasher_SOURCES = trap-crasher.c trap_crasher_OBJECTS = trap-crasher.$(OBJEXT) trap_crasher_LDADD = $(LDADD) @@ -310,26 +351,30 @@ AM_V_CCLD = $(am__v_CCLD_@AM_V@) am__v_CCLD_ = $(am__v_CCLD_@AM_DEFAULT_V@) am__v_CCLD_0 = @echo " CCLD " $@; am__v_CCLD_1 = -SOURCES = $(libutils_la_SOURCES) a1-trap-test.c affine-test.c \ - alpha-loop.c alphamap.c blitters-test.c check-formats.c \ - combiner-test.c composite.c composite-traps-test.c \ +SOURCES = $(libutils_la_SOURCES) a1-trap-test.c affine-bench.c \ + affine-test.c alpha-loop.c alphamap.c blitters-test.c \ + check-formats.c combiner-test.c composite.c \ + composite-traps-test.c cover-test.c fence-image-self-test.c \ fetch-test.c glyph-test.c gradient-crash-test.c \ infinite-loop.c lowlevel-blt-bench.c matrix-test.c oob-test.c \ - pdf-op-test.c pixel-test.c prng-test.c radial-perf-test.c \ - region-contains-test.c region-test.c region-translate-test.c \ - rotate-test.c scaling-bench.c scaling-crash-test.c \ - scaling-helpers-test.c scaling-test.c stress-test.c \ - thread-test.c trap-crasher.c -DIST_SOURCES = $(libutils_la_SOURCES) a1-trap-test.c affine-test.c \ - alpha-loop.c alphamap.c blitters-test.c check-formats.c \ - combiner-test.c composite.c composite-traps-test.c \ + pdf-op-test.c pixel-test.c prng-test.c radial-invalid.c \ + radial-perf-test.c region-contains-test.c region-test.c \ + region-translate-test.c rotate-test.c scaling-bench.c \ + scaling-crash-test.c scaling-helpers-test.c scaling-test.c \ + solid-test.c stress-test.c thread-test.c tolerance-test.c \ + trap-crasher.c +DIST_SOURCES = $(libutils_la_SOURCES) a1-trap-test.c affine-bench.c \ + affine-test.c alpha-loop.c alphamap.c blitters-test.c \ + check-formats.c combiner-test.c composite.c \ + composite-traps-test.c cover-test.c fence-image-self-test.c \ fetch-test.c glyph-test.c gradient-crash-test.c \ infinite-loop.c lowlevel-blt-bench.c matrix-test.c oob-test.c \ - pdf-op-test.c pixel-test.c prng-test.c radial-perf-test.c \ - region-contains-test.c region-test.c region-translate-test.c \ - rotate-test.c scaling-bench.c scaling-crash-test.c \ - scaling-helpers-test.c scaling-test.c stress-test.c \ - thread-test.c trap-crasher.c + pdf-op-test.c pixel-test.c prng-test.c radial-invalid.c \ + radial-perf-test.c region-contains-test.c region-test.c \ + region-translate-test.c rotate-test.c scaling-bench.c \ + scaling-crash-test.c scaling-helpers-test.c scaling-test.c \ + solid-test.c stress-test.c thread-test.c tolerance-test.c \ + trap-crasher.c am__can_run_installinfo = \ case $$AM_UPDATE_INFO_DIR in \ n|no|NO) false;; \ @@ -491,33 +536,38 @@ top_srcdir = @top_srcdir@ # Tests (sorted by expected completion time) TESTPROGRAMS = \ - prng-test \ - a1-trap-test \ - pdf-op-test \ - region-test \ - region-translate-test \ - combiner-test \ - pixel-test \ - fetch-test \ - rotate-test \ - oob-test \ - infinite-loop \ - trap-crasher \ - alpha-loop \ - thread-test \ - scaling-crash-test \ - scaling-helpers-test \ - gradient-crash-test \ - region-contains-test \ - alphamap \ - matrix-test \ - stress-test \ - composite-traps-test \ - blitters-test \ - glyph-test \ - scaling-test \ - affine-test \ - composite \ + oob-test \ + infinite-loop \ + trap-crasher \ + fence-image-self-test \ + region-translate-test \ + fetch-test \ + a1-trap-test \ + prng-test \ + radial-invalid \ + pdf-op-test \ + region-test \ + combiner-test \ + scaling-crash-test \ + alpha-loop \ + scaling-helpers-test \ + thread-test \ + rotate-test \ + alphamap \ + gradient-crash-test \ + pixel-test \ + matrix-test \ + composite-traps-test \ + region-contains-test \ + glyph-test \ + solid-test \ + stress-test \ + cover-test \ + blitters-test \ + affine-test \ + scaling-test \ + composite \ + tolerance-test \ $(NULL) @@ -527,6 +577,7 @@ OTHERPROGRAMS = \ radial-perf-test \ check-formats \ scaling-bench \ + affine-bench \ $(NULL) @@ -607,6 +658,9 @@ clean-noinstPROGRAMS: a1-trap-test$(EXEEXT): $(a1_trap_test_OBJECTS) $(a1_trap_test_DEPENDENCIES) $(EXTRA_a1_trap_test_DEPENDENCIES) @rm -f a1-trap-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(a1_trap_test_OBJECTS) $(a1_trap_test_LDADD) $(LIBS) +affine-bench$(EXEEXT): $(affine_bench_OBJECTS) $(affine_bench_DEPENDENCIES) $(EXTRA_affine_bench_DEPENDENCIES) + @rm -f affine-bench$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(affine_bench_OBJECTS) $(affine_bench_LDADD) $(LIBS) affine-test$(EXEEXT): $(affine_test_OBJECTS) $(affine_test_DEPENDENCIES) $(EXTRA_affine_test_DEPENDENCIES) @rm -f affine-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(affine_test_OBJECTS) $(affine_test_LDADD) $(LIBS) @@ -631,6 +685,12 @@ composite$(EXEEXT): $(composite_OBJECTS) $(composite_DEPENDENCIES) $(EXTRA_compo composite-traps-test$(EXEEXT): $(composite_traps_test_OBJECTS) $(composite_traps_test_DEPENDENCIES) $(EXTRA_composite_traps_test_DEPENDENCIES) @rm -f composite-traps-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(composite_traps_test_OBJECTS) $(composite_traps_test_LDADD) $(LIBS) +cover-test$(EXEEXT): $(cover_test_OBJECTS) $(cover_test_DEPENDENCIES) $(EXTRA_cover_test_DEPENDENCIES) + @rm -f cover-test$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(cover_test_OBJECTS) $(cover_test_LDADD) $(LIBS) +fence-image-self-test$(EXEEXT): $(fence_image_self_test_OBJECTS) $(fence_image_self_test_DEPENDENCIES) $(EXTRA_fence_image_self_test_DEPENDENCIES) + @rm -f fence-image-self-test$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(fence_image_self_test_OBJECTS) $(fence_image_self_test_LDADD) $(LIBS) fetch-test$(EXEEXT): $(fetch_test_OBJECTS) $(fetch_test_DEPENDENCIES) $(EXTRA_fetch_test_DEPENDENCIES) @rm -f fetch-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(fetch_test_OBJECTS) $(fetch_test_LDADD) $(LIBS) @@ -661,6 +721,9 @@ pixel-test$(EXEEXT): $(pixel_test_OBJECTS) $(pixel_test_DEPENDENCIES) $(EXTRA_pi prng-test$(EXEEXT): $(prng_test_OBJECTS) $(prng_test_DEPENDENCIES) $(EXTRA_prng_test_DEPENDENCIES) @rm -f prng-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(prng_test_OBJECTS) $(prng_test_LDADD) $(LIBS) +radial-invalid$(EXEEXT): $(radial_invalid_OBJECTS) $(radial_invalid_DEPENDENCIES) $(EXTRA_radial_invalid_DEPENDENCIES) + @rm -f radial-invalid$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(radial_invalid_OBJECTS) $(radial_invalid_LDADD) $(LIBS) radial-perf-test$(EXEEXT): $(radial_perf_test_OBJECTS) $(radial_perf_test_DEPENDENCIES) $(EXTRA_radial_perf_test_DEPENDENCIES) @rm -f radial-perf-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(radial_perf_test_OBJECTS) $(radial_perf_test_LDADD) $(LIBS) @@ -688,12 +751,18 @@ scaling-helpers-test$(EXEEXT): $(scaling_helpers_test_OBJECTS) $(scaling_helpers scaling-test$(EXEEXT): $(scaling_test_OBJECTS) $(scaling_test_DEPENDENCIES) $(EXTRA_scaling_test_DEPENDENCIES) @rm -f scaling-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(scaling_test_OBJECTS) $(scaling_test_LDADD) $(LIBS) +solid-test$(EXEEXT): $(solid_test_OBJECTS) $(solid_test_DEPENDENCIES) $(EXTRA_solid_test_DEPENDENCIES) + @rm -f solid-test$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(solid_test_OBJECTS) $(solid_test_LDADD) $(LIBS) stress-test$(EXEEXT): $(stress_test_OBJECTS) $(stress_test_DEPENDENCIES) $(EXTRA_stress_test_DEPENDENCIES) @rm -f stress-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(stress_test_OBJECTS) $(stress_test_LDADD) $(LIBS) thread-test$(EXEEXT): $(thread_test_OBJECTS) $(thread_test_DEPENDENCIES) $(EXTRA_thread_test_DEPENDENCIES) @rm -f thread-test$(EXEEXT) $(AM_V_CCLD)$(LINK) $(thread_test_OBJECTS) $(thread_test_LDADD) $(LIBS) +tolerance-test$(EXEEXT): $(tolerance_test_OBJECTS) $(tolerance_test_DEPENDENCIES) $(EXTRA_tolerance_test_DEPENDENCIES) + @rm -f tolerance-test$(EXEEXT) + $(AM_V_CCLD)$(LINK) $(tolerance_test_OBJECTS) $(tolerance_test_LDADD) $(LIBS) trap-crasher$(EXEEXT): $(trap_crasher_OBJECTS) $(trap_crasher_DEPENDENCIES) $(EXTRA_trap_crasher_DEPENDENCIES) @rm -f trap-crasher$(EXEEXT) $(AM_V_CCLD)$(LINK) $(trap_crasher_OBJECTS) $(trap_crasher_LDADD) $(LIBS) @@ -705,6 +774,7 @@ distclean-compile: -rm -f *.tab.c @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/a1-trap-test.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/affine-bench.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/affine-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alpha-loop.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alphamap.Po@am__quote@ @@ -713,6 +783,8 @@ distclean-compile: @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/combiner-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/composite-traps-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/composite.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/cover-test.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fence-image-self-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fetch-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/glyph-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/gradient-crash-test.Po@am__quote@ @@ -723,6 +795,7 @@ distclean-compile: @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pdf-op-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixel-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/prng-test.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/radial-invalid.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/radial-perf-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/region-contains-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/region-test.Po@am__quote@ @@ -732,8 +805,10 @@ distclean-compile: @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/scaling-crash-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/scaling-helpers-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/scaling-test.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/solid-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/stress-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/thread-test.Po@am__quote@ +@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tolerance-test.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/trap-crasher.Po@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/utils-prng.Plo@am__quote@ @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/utils.Plo@am__quote@ diff --git a/lib/pixman/test/Makefile.sources b/lib/pixman/test/Makefile.sources index 2ae5d9f8d..5d55e676a 100644 --- a/lib/pixman/test/Makefile.sources +++ b/lib/pixman/test/Makefile.sources @@ -1,32 +1,37 @@ # Tests (sorted by expected completion time) -TESTPROGRAMS = \ - prng-test \ - a1-trap-test \ - pdf-op-test \ - region-test \ - region-translate-test \ - combiner-test \ - pixel-test \ - fetch-test \ - rotate-test \ - oob-test \ - infinite-loop \ - trap-crasher \ - alpha-loop \ - thread-test \ - scaling-crash-test \ - scaling-helpers-test \ - gradient-crash-test \ - region-contains-test \ - alphamap \ - matrix-test \ - stress-test \ - composite-traps-test \ - blitters-test \ - glyph-test \ - scaling-test \ - affine-test \ - composite \ +TESTPROGRAMS = \ + oob-test \ + infinite-loop \ + trap-crasher \ + fence-image-self-test \ + region-translate-test \ + fetch-test \ + a1-trap-test \ + prng-test \ + radial-invalid \ + pdf-op-test \ + region-test \ + combiner-test \ + scaling-crash-test \ + alpha-loop \ + scaling-helpers-test \ + thread-test \ + rotate-test \ + alphamap \ + gradient-crash-test \ + pixel-test \ + matrix-test \ + composite-traps-test \ + region-contains-test \ + glyph-test \ + solid-test \ + stress-test \ + cover-test \ + blitters-test \ + affine-test \ + scaling-test \ + composite \ + tolerance-test \ $(NULL) # Other programs @@ -35,6 +40,7 @@ OTHERPROGRAMS = \ radial-perf-test \ check-formats \ scaling-bench \ + affine-bench \ $(NULL) # Utility functions diff --git a/lib/pixman/test/affine-bench.c b/lib/pixman/test/affine-bench.c new file mode 100644 index 000000000..86bf46ef7 --- /dev/null +++ b/lib/pixman/test/affine-bench.c @@ -0,0 +1,448 @@ +/* + * Copyright © 2014 RISC OS Open Ltd + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. The copyright holders make no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Ben Avison (bavison@riscosopen.org) + */ + +#include +#include +#include +#include +#include +#include "utils.h" + +#ifdef HAVE_GETTIMEOFDAY +#include +#else +#include +#endif + +#define WIDTH 1920 +#define HEIGHT 1080 + +/* How much data to read to flush all cached data to RAM */ +#define MAX_L2CACHE_SIZE (8 * 1024 * 1024) + +#define PAGE_SIZE (4 * 1024) + +struct bench_info +{ + pixman_op_t op; + pixman_transform_t transform; + pixman_image_t *src_image; + pixman_image_t *mask_image; + pixman_image_t *dest_image; + int32_t src_x; + int32_t src_y; +}; + +typedef struct bench_info bench_info_t; + +struct box_48_16 +{ + pixman_fixed_48_16_t x1; + pixman_fixed_48_16_t y1; + pixman_fixed_48_16_t x2; + pixman_fixed_48_16_t y2; +}; + +typedef struct box_48_16 box_48_16_t; + +/* This function is copied verbatim from pixman.c. */ +static pixman_bool_t +compute_transformed_extents (pixman_transform_t *transform, + const pixman_box32_t *extents, + box_48_16_t *transformed) +{ + pixman_fixed_48_16_t tx1, ty1, tx2, ty2; + pixman_fixed_t x1, y1, x2, y2; + int i; + + x1 = pixman_int_to_fixed (extents->x1) + pixman_fixed_1 / 2; + y1 = pixman_int_to_fixed (extents->y1) + pixman_fixed_1 / 2; + x2 = pixman_int_to_fixed (extents->x2) - pixman_fixed_1 / 2; + y2 = pixman_int_to_fixed (extents->y2) - pixman_fixed_1 / 2; + + if (!transform) + { + transformed->x1 = x1; + transformed->y1 = y1; + transformed->x2 = x2; + transformed->y2 = y2; + + return TRUE; + } + + tx1 = ty1 = INT64_MAX; + tx2 = ty2 = INT64_MIN; + + for (i = 0; i < 4; ++i) + { + pixman_fixed_48_16_t tx, ty; + pixman_vector_t v; + + v.vector[0] = (i & 0x01)? x1 : x2; + v.vector[1] = (i & 0x02)? y1 : y2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point (transform, &v)) + return FALSE; + + tx = (pixman_fixed_48_16_t)v.vector[0]; + ty = (pixman_fixed_48_16_t)v.vector[1]; + + if (tx < tx1) + tx1 = tx; + if (ty < ty1) + ty1 = ty; + if (tx > tx2) + tx2 = tx; + if (ty > ty2) + ty2 = ty; + } + + transformed->x1 = tx1; + transformed->y1 = ty1; + transformed->x2 = tx2; + transformed->y2 = ty2; + + return TRUE; +} + +static void +create_image (uint32_t width, + uint32_t height, + pixman_format_code_t format, + pixman_filter_t filter, + uint32_t **bits, + pixman_image_t **image) +{ + uint32_t stride = (width * PIXMAN_FORMAT_BPP (format) + 31) / 32 * 4; + + *bits = aligned_malloc (PAGE_SIZE, stride * height); + memset (*bits, 0xCC, stride * height); + *image = pixman_image_create_bits (format, width, height, *bits, stride); + pixman_image_set_repeat (*image, PIXMAN_REPEAT_NORMAL); + pixman_image_set_filter (*image, filter, NULL, 0); +} + +/* This needs to match the shortest cacheline length we expect to encounter */ +#define CACHE_CLEAN_INCREMENT 32 + +static void +flush_cache (void) +{ + static const char clean_space[MAX_L2CACHE_SIZE]; + volatile const char *x = clean_space; + const char *clean_end = clean_space + sizeof clean_space; + + while (x < clean_end) + { + (void) *x; + x += CACHE_CLEAN_INCREMENT; + } +} + +/* Obtain current time in microseconds modulo 2^32 */ +uint32_t +gettimei (void) +{ +#ifdef HAVE_GETTIMEOFDAY + struct timeval tv; + + gettimeofday (&tv, NULL); + return tv.tv_sec * 1000000 + tv.tv_usec; +#else + return (uint64_t) clock () * 1000000 / CLOCKS_PER_SEC; +#endif +} + +static void +pixman_image_composite_wrapper (const pixman_composite_info_t *info) +{ + pixman_image_composite (info->op, + info->src_image, info->mask_image, info->dest_image, + info->src_x, info->src_y, + info->mask_x, info->mask_y, + info->dest_x, info->dest_y, + info->width, info->height); +} + +static void +pixman_image_composite_empty (const pixman_composite_info_t *info) +{ + pixman_image_composite (info->op, + info->src_image, info->mask_image, info->dest_image, + info->src_x, info->src_y, + info->mask_x, info->mask_y, + info->dest_x, info->dest_y, + 1, 1); +} + +static void +bench (const bench_info_t *bi, + uint32_t max_n, + uint32_t max_time, + uint32_t *ret_n, + uint32_t *ret_time, + void (*func) (const pixman_composite_info_t *info)) +{ + uint32_t n = 0; + uint32_t t0; + uint32_t t1; + uint32_t x = 0; + pixman_transform_t t; + pixman_composite_info_t info; + + t = bi->transform; + info.op = bi->op; + info.src_image = bi->src_image; + info.mask_image = bi->mask_image; + info.dest_image = bi->dest_image; + info.src_x = 0; + info.src_y = 0; + info.mask_x = 0; + info.mask_y = 0; + /* info.dest_x set below */ + info.dest_y = 0; + info.width = WIDTH; + info.height = HEIGHT; + + t0 = gettimei (); + + do + { + + if (++x >= 64) + x = 0; + + info.dest_x = 63 - x; + + t.matrix[0][2] = pixman_int_to_fixed (bi->src_x + x); + t.matrix[1][2] = pixman_int_to_fixed (bi->src_y); + pixman_image_set_transform (bi->src_image, &t); + + if (bi->mask_image) + pixman_image_set_transform (bi->mask_image, &t); + + func (&info); + t1 = gettimei (); + } + while (++n < max_n && (t1 - t0) < max_time); + + if (ret_n) + *ret_n = n; + + *ret_time = t1 - t0; +} + +int +parse_fixed_argument (char *arg, pixman_fixed_t *value) +{ + char *tailptr; + + *value = pixman_double_to_fixed (strtod (arg, &tailptr)); + + return *tailptr == '\0'; +} + +int +parse_arguments (int argc, + char *argv[], + pixman_transform_t *t, + pixman_op_t *op, + pixman_format_code_t *src_format, + pixman_format_code_t *mask_format, + pixman_format_code_t *dest_format) +{ + if (!parse_fixed_argument (*argv, &t->matrix[0][0])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[0][1])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[1][0])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[1][1])) + return 0; + + if (*++argv == NULL) + return 1; + + *op = operator_from_string (*argv); + if (*op == PIXMAN_OP_NONE) + return 0; + + if (*++argv == NULL) + return 1; + + *src_format = format_from_string (*argv); + if (*src_format == PIXMAN_null) + return 0; + + ++argv; + if (argv[0] && argv[1]) + { + *mask_format = format_from_string (*argv); + if (*mask_format == PIXMAN_null) + return 0; + ++argv; + } + if (*argv) + { + *dest_format = format_from_string (*argv); + if (*dest_format == PIXMAN_null) + return 0; + } + return 1; +} + +static void +run_benchmark (const bench_info_t *bi) +{ + uint32_t n; /* number of iterations in at least 5 seconds */ + uint32_t t1; /* time taken to do n iterations, microseconds */ + uint32_t t2; /* calling overhead for n iterations, microseconds */ + + flush_cache (); + bench (bi, UINT32_MAX, 5000000, &n, &t1, pixman_image_composite_wrapper); + bench (bi, n, UINT32_MAX, NULL, &t2, pixman_image_composite_empty); + + /* The result indicates the output rate in megapixels/second */ + printf ("%6.2f\n", (double) n * WIDTH * HEIGHT / (t1 - t2)); +} + + +int +main (int argc, char *argv[]) +{ + bench_info_t binfo; + pixman_filter_t filter = PIXMAN_FILTER_NEAREST; + pixman_format_code_t src_format = PIXMAN_a8r8g8b8; + pixman_format_code_t mask_format = 0; + pixman_format_code_t dest_format = PIXMAN_a8r8g8b8; + pixman_box32_t dest_box = { 0, 0, WIDTH, HEIGHT }; + box_48_16_t transformed = { 0 }; + int32_t xmin, ymin, xmax, ymax; + uint32_t *src, *mask, *dest; + + binfo.op = PIXMAN_OP_SRC; + binfo.mask_image = NULL; + pixman_transform_init_identity (&binfo.transform); + + ++argv; + if (*argv && (*argv)[0] == '-' && (*argv)[1] == 'n') + { + filter = PIXMAN_FILTER_NEAREST; + ++argv; + --argc; + } + + if (*argv && (*argv)[0] == '-' && (*argv)[1] == 'b') + { + filter = PIXMAN_FILTER_BILINEAR; + ++argv; + --argc; + } + + if (argc == 1 || + !parse_arguments (argc, argv, &binfo.transform, &binfo.op, + &src_format, &mask_format, &dest_format)) + { + printf ("Usage: affine-bench [-n] [-b] axx [axy] [ayx] [ayy] [combine type]\n"); + printf (" [src format] [mask format] [dest format]\n"); + printf (" -n : nearest scaling (default)\n"); + printf (" -b : bilinear scaling\n"); + printf (" axx : x_out:x_in factor\n"); + printf (" axy : x_out:y_in factor (default 0)\n"); + printf (" ayx : y_out:x_in factor (default 0)\n"); + printf (" ayy : y_out:y_in factor (default 1)\n"); + printf (" combine type : src, over, in etc (default src)\n"); + printf (" src format : a8r8g8b8, r5g6b5 etc (default a8r8g8b8)\n"); + printf (" mask format : as for src format, but no mask used if omitted\n"); + printf (" dest format : as for src format (default a8r8g8b8)\n"); + printf ("The output is a single number in megapixels/second.\n"); + + return EXIT_FAILURE; + } + + /* Compute required extents for source and mask image so they qualify + * for COVER fast paths and get the flags in pixman.c:analyze_extent(). + * These computations are for FAST_PATH_SAMPLES_COVER_CLIP_BILINEAR, + * but at the same time they also allow COVER_CLIP_NEAREST. + */ + compute_transformed_extents (&binfo.transform, &dest_box, &transformed); + xmin = pixman_fixed_to_int (transformed.x1 - pixman_fixed_1 / 2); + ymin = pixman_fixed_to_int (transformed.y1 - pixman_fixed_1 / 2); + xmax = pixman_fixed_to_int (transformed.x2 + pixman_fixed_1 / 2); + ymax = pixman_fixed_to_int (transformed.y2 + pixman_fixed_1 / 2); + /* Note: + * The upper limits can be reduced to the following when fetchers + * are guaranteed to not access pixels with zero weight. This concerns + * particularly all bilinear samplers. + * + * xmax = pixman_fixed_to_int (transformed.x2 + pixman_fixed_1 / 2 - pixman_fixed_e); + * ymax = pixman_fixed_to_int (transformed.y2 + pixman_fixed_1 / 2 - pixman_fixed_e); + * This is equivalent to subtracting 0.5 and rounding up, rather than + * subtracting 0.5, rounding down and adding 1. + */ + binfo.src_x = -xmin; + binfo.src_y = -ymin; + + /* Always over-allocate width by 64 pixels for all src, mask and dst, + * so that we can iterate over an x-offset 0..63 in bench (). + * This is similar to lowlevel-blt-bench, which uses the same method + * to hit different cacheline misalignments. + */ + create_image (xmax - xmin + 64, ymax - ymin + 1, src_format, filter, + &src, &binfo.src_image); + + if (mask_format) + { + create_image (xmax - xmin + 64, ymax - ymin + 1, mask_format, filter, + &mask, &binfo.mask_image); + + if ((PIXMAN_FORMAT_R(mask_format) || + PIXMAN_FORMAT_G(mask_format) || + PIXMAN_FORMAT_B(mask_format))) + { + pixman_image_set_component_alpha (binfo.mask_image, 1); + } + } + + create_image (WIDTH + 64, HEIGHT, dest_format, filter, + &dest, &binfo.dest_image); + + run_benchmark (&binfo); + + return EXIT_SUCCESS; +} diff --git a/lib/pixman/test/blitters-test.c b/lib/pixman/test/blitters-test.c index ea03f475d..026f4b006 100644 --- a/lib/pixman/test/blitters-test.c +++ b/lib/pixman/test/blitters-test.c @@ -122,6 +122,15 @@ static pixman_op_t op_list[] = { PIXMAN_OP_ATOP_REVERSE, PIXMAN_OP_XOR, PIXMAN_OP_ADD, + PIXMAN_OP_MULTIPLY, + PIXMAN_OP_SCREEN, + PIXMAN_OP_OVERLAY, + PIXMAN_OP_DARKEN, + PIXMAN_OP_LIGHTEN, + PIXMAN_OP_HARD_LIGHT, + PIXMAN_OP_DIFFERENCE, + PIXMAN_OP_EXCLUSION, +#if 0 /* these use floating point math and are not always bitexact on different platforms */ PIXMAN_OP_SATURATE, PIXMAN_OP_DISJOINT_CLEAR, PIXMAN_OP_DISJOINT_SRC, @@ -147,17 +156,8 @@ static pixman_op_t op_list[] = { PIXMAN_OP_CONJOINT_ATOP, PIXMAN_OP_CONJOINT_ATOP_REVERSE, PIXMAN_OP_CONJOINT_XOR, - PIXMAN_OP_MULTIPLY, - PIXMAN_OP_SCREEN, - PIXMAN_OP_OVERLAY, - PIXMAN_OP_DARKEN, - PIXMAN_OP_LIGHTEN, PIXMAN_OP_COLOR_DODGE, PIXMAN_OP_COLOR_BURN, - PIXMAN_OP_HARD_LIGHT, - PIXMAN_OP_DIFFERENCE, - PIXMAN_OP_EXCLUSION, -#if 0 /* these use floating point math and are not always bitexact on different platforms */ PIXMAN_OP_SOFT_LIGHT, PIXMAN_OP_HSL_HUE, PIXMAN_OP_HSL_SATURATION, @@ -394,6 +394,6 @@ main (int argc, const char *argv[]) } return fuzzer_test_main("blitters", 2000000, - 0xE0A07495, + 0xCC21DDF0, test_composite, argc, argv); } diff --git a/lib/pixman/test/check-formats.c b/lib/pixman/test/check-formats.c index 7edc198c1..4e2633c41 100644 --- a/lib/pixman/test/check-formats.c +++ b/lib/pixman/test/check-formats.c @@ -104,182 +104,6 @@ check_op (pixman_op_t op, return retval; } -static const pixman_op_t op_list[] = -{ - PIXMAN_OP_CLEAR, - PIXMAN_OP_SRC, - PIXMAN_OP_DST, - PIXMAN_OP_OVER, - PIXMAN_OP_OVER_REVERSE, - PIXMAN_OP_IN, - PIXMAN_OP_IN_REVERSE, - PIXMAN_OP_OUT, - PIXMAN_OP_OUT_REVERSE, - PIXMAN_OP_ATOP, - PIXMAN_OP_ATOP_REVERSE, - PIXMAN_OP_XOR, - PIXMAN_OP_ADD, - PIXMAN_OP_SATURATE, - - PIXMAN_OP_DISJOINT_CLEAR, - PIXMAN_OP_DISJOINT_SRC, - PIXMAN_OP_DISJOINT_DST, - PIXMAN_OP_DISJOINT_OVER, - PIXMAN_OP_DISJOINT_OVER_REVERSE, - PIXMAN_OP_DISJOINT_IN, - PIXMAN_OP_DISJOINT_IN_REVERSE, - PIXMAN_OP_DISJOINT_OUT, - PIXMAN_OP_DISJOINT_OUT_REVERSE, - PIXMAN_OP_DISJOINT_ATOP, - PIXMAN_OP_DISJOINT_ATOP_REVERSE, - PIXMAN_OP_DISJOINT_XOR, - - PIXMAN_OP_CONJOINT_CLEAR, - PIXMAN_OP_CONJOINT_SRC, - PIXMAN_OP_CONJOINT_DST, - PIXMAN_OP_CONJOINT_OVER, - PIXMAN_OP_CONJOINT_OVER_REVERSE, - PIXMAN_OP_CONJOINT_IN, - PIXMAN_OP_CONJOINT_IN_REVERSE, - PIXMAN_OP_CONJOINT_OUT, - PIXMAN_OP_CONJOINT_OUT_REVERSE, - PIXMAN_OP_CONJOINT_ATOP, - PIXMAN_OP_CONJOINT_ATOP_REVERSE, - PIXMAN_OP_CONJOINT_XOR, -}; - -static const pixman_format_code_t format_list[] = -{ - PIXMAN_a8r8g8b8, - PIXMAN_x8r8g8b8, - PIXMAN_a8b8g8r8, - PIXMAN_x8b8g8r8, - PIXMAN_b8g8r8a8, - PIXMAN_b8g8r8x8, - PIXMAN_r8g8b8a8, - PIXMAN_r8g8b8x8, - PIXMAN_x14r6g6b6, - PIXMAN_x2r10g10b10, - PIXMAN_a2r10g10b10, - PIXMAN_x2b10g10r10, - PIXMAN_a2b10g10r10, - PIXMAN_a8r8g8b8_sRGB, - PIXMAN_r8g8b8, - PIXMAN_b8g8r8, - PIXMAN_r5g6b5, - PIXMAN_b5g6r5, - PIXMAN_a1r5g5b5, - PIXMAN_x1r5g5b5, - PIXMAN_a1b5g5r5, - PIXMAN_x1b5g5r5, - PIXMAN_a4r4g4b4, - PIXMAN_x4r4g4b4, - PIXMAN_a4b4g4r4, - PIXMAN_x4b4g4r4, - PIXMAN_a8, - PIXMAN_r3g3b2, - PIXMAN_b2g3r3, - PIXMAN_a2r2g2b2, - PIXMAN_a2b2g2r2, - PIXMAN_x4a4, - PIXMAN_a4, - PIXMAN_r1g2b1, - PIXMAN_b1g2r1, - PIXMAN_a1r1g1b1, - PIXMAN_a1b1g1r1, - PIXMAN_a1, -}; - -static pixman_format_code_t -format_from_string (const char *s) -{ - int i; - - for (i = 0; i < ARRAY_LENGTH (format_list); ++i) - { - if (strcasecmp (format_name (format_list[i]), s) == 0) - return format_list[i]; - } - - return PIXMAN_null; -} - -static void -emit (const char *s, int *n_chars) -{ - *n_chars += printf ("%s,", s); - if (*n_chars > 60) - { - printf ("\n "); - *n_chars = 0; - } - else - { - printf (" "); - (*n_chars)++; - } -} - -static void -list_formats (void) -{ - int n_chars; - int i; - - printf ("Formats:\n "); - - n_chars = 0; - for (i = 0; i < ARRAY_LENGTH (format_list); ++i) - emit (format_name (format_list[i]), &n_chars); - - printf ("\n\n"); -} - -static void -list_operators (void) -{ - char short_name [128] = { 0 }; - int i, n_chars; - - printf ("Operators:\n "); - - n_chars = 0; - for (i = 0; i < ARRAY_LENGTH (op_list); ++i) - { - pixman_op_t op = op_list[i]; - int j; - - snprintf (short_name, sizeof (short_name) - 1, "%s", - operator_name (op) + strlen ("PIXMAN_OP_")); - - for (j = 0; short_name[j] != '\0'; ++j) - short_name[j] = tolower (short_name[j]); - - emit (short_name, &n_chars); - } - - printf ("\n\n"); -} - -static pixman_op_t -operator_from_string (const char *s) -{ - char full_name[128] = { 0 }; - int i; - - snprintf (full_name, (sizeof full_name) - 1, "PIXMAN_OP_%s", s); - - for (i = 0; i < ARRAY_LENGTH (op_list); ++i) - { - pixman_op_t op = op_list[i]; - - if (strcasecmp (operator_name (op), full_name) == 0) - return op; - } - - return PIXMAN_OP_NONE; -} - int main (int argc, char **argv) { diff --git a/lib/pixman/test/composite.c b/lib/pixman/test/composite.c index 9e51a8f65..594c69772 100644 --- a/lib/pixman/test/composite.c +++ b/lib/pixman/test/composite.c @@ -299,17 +299,6 @@ composite_test (image_t *dst, } } - if (mask) - { - if (component_alpha && PIXMAN_FORMAT_R (mask->format) == 0) - { - /* Ax component-alpha masks expand alpha into - * all color channels. - */ - tmsk.r = tmsk.g = tmsk.b = tmsk.a; - } - } - if (PIXMAN_FORMAT_TYPE (dst->format) == PIXMAN_TYPE_ARGB_SRGB) { tdst.r = convert_linear_to_srgb (tdst.r); diff --git a/lib/pixman/test/cover-test.c b/lib/pixman/test/cover-test.c new file mode 100644 index 000000000..83e2972d9 --- /dev/null +++ b/lib/pixman/test/cover-test.c @@ -0,0 +1,449 @@ +/* + * Copyright © 2015 RISC OS Open Ltd + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. The copyright holders make no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Ben Avison (bavison@riscosopen.org) + * + */ + +/* + * This test aims to verify both numerical correctness and the honouring of + * array bounds for scaled plots (both nearest-neighbour and bilinear) at or + * close to the boundary conditions for applicability of "cover" type fast paths + * and iter fetch routines. + * + * It has a secondary purpose: by setting the env var EXACT (to any value) it + * will only test plots that are exactly on the boundary condition. This makes + * it possible to ensure that "cover" routines are being used to the maximum, + * although this requires the use of a debugger or code instrumentation to + * verify. + */ + +#include "utils.h" +#include +#include + +/* Approximate limits for random scale factor generation - these ensure we can + * get at least 8x reduction and 8x enlargement. + */ +#define LOG2_MAX_FACTOR (3) + +/* 1/sqrt(2) (or sqrt(0.5), or 2^-0.5) as a 0.32 fixed-point number */ +#define INV_SQRT_2_0POINT32_FIXED (0xB504F334u) + +/* The largest increment that can be generated by random_scale_factor(). + * This occurs when the "mantissa" part is 0xFFFFFFFF and the "exponent" + * part is -LOG2_MAX_FACTOR. + */ +#define MAX_INC ((pixman_fixed_t) \ + (INV_SQRT_2_0POINT32_FIXED >> (31 - 16 - LOG2_MAX_FACTOR))) + +/* Minimum source width (in pixels) based on a typical page size of 4K and + * maximum colour depth of 32bpp. + */ +#define MIN_SRC_WIDTH (4096 / 4) + +/* Derive the destination width so that at max increment we fit within source */ +#define DST_WIDTH (MIN_SRC_WIDTH * pixman_fixed_1 / MAX_INC) + +/* Calculate heights the other way round. + * No limits due to page alignment here. + */ +#define DST_HEIGHT 3 +#define SRC_HEIGHT ((DST_HEIGHT * MAX_INC + pixman_fixed_1 - 1) / pixman_fixed_1) + +/* At the time of writing, all the scaled fast paths use SRC, OVER or ADD + * Porter-Duff operators. XOR is included in the list to ensure good + * representation of iter scanline fetch routines. + */ +static const pixman_op_t op_list[] = { + PIXMAN_OP_SRC, + PIXMAN_OP_OVER, + PIXMAN_OP_ADD, + PIXMAN_OP_XOR, +}; + +/* At the time of writing, all the scaled fast paths use a8r8g8b8, x8r8g8b8 + * or r5g6b5, or red-blue swapped versions of the same. When a mask channel is + * used, it is always a8 (and so implicitly not component alpha). a1r5g5b5 is + * included because it is the only other format to feature in any iters. */ +static const pixman_format_code_t img_fmt_list[] = { + PIXMAN_a8r8g8b8, + PIXMAN_x8r8g8b8, + PIXMAN_r5g6b5, + PIXMAN_a1r5g5b5 +}; + +/* This is a flag reflecting the environment variable EXACT. It can be used + * to ensure that source coordinates corresponding exactly to the "cover" limits + * are used, rather than any "near misses". This can, for example, be used in + * conjunction with a debugger to ensure that only COVER fast paths are used. + */ +static int exact; + +static pixman_image_t * +create_src_image (pixman_format_code_t fmt) +{ + pixman_image_t *tmp_img, *img; + + /* We need the left-most and right-most MIN_SRC_WIDTH pixels to have + * predictable values, even though fence_image_create_bits() may allocate + * an image somewhat larger than that, by an amount that varies depending + * upon the page size on the current platform. The solution is to create a + * temporary non-fenced image that is exactly MIN_SRC_WIDTH wide and blit it + * into the fenced image. + */ + tmp_img = pixman_image_create_bits (fmt, MIN_SRC_WIDTH, SRC_HEIGHT, + NULL, 0); + if (tmp_img == NULL) + return NULL; + + img = fence_image_create_bits (fmt, MIN_SRC_WIDTH, SRC_HEIGHT, TRUE); + if (img == NULL) + { + pixman_image_unref (tmp_img); + return NULL; + } + + prng_randmemset (tmp_img->bits.bits, + tmp_img->bits.rowstride * SRC_HEIGHT * sizeof (uint32_t), + 0); + image_endian_swap (tmp_img); + + pixman_image_composite (PIXMAN_OP_SRC, tmp_img, NULL, img, + 0, 0, 0, 0, 0, 0, + MIN_SRC_WIDTH, SRC_HEIGHT); + pixman_image_composite (PIXMAN_OP_SRC, tmp_img, NULL, img, + 0, 0, 0, 0, img->bits.width - MIN_SRC_WIDTH, 0, + MIN_SRC_WIDTH, SRC_HEIGHT); + + pixman_image_unref (tmp_img); + + return img; +} + +static pixman_fixed_t +random_scale_factor(void) +{ + /* Get a random number with top bit set. */ + uint32_t f = prng_rand () | 0x80000000u; + + /* In log(2) space, this is still approximately evenly spread between 31 + * and 32. Divide by sqrt(2) to centre the distribution on 2^31. + */ + f = ((uint64_t) f * INV_SQRT_2_0POINT32_FIXED) >> 32; + + /* Now shift right (ie divide by an integer power of 2) to spread the + * distribution between centres at 2^(16 +/- LOG2_MAX_FACTOR). + */ + f >>= 31 - 16 + prng_rand_n (2 * LOG2_MAX_FACTOR + 1) - LOG2_MAX_FACTOR; + + return f; +} + +static pixman_fixed_t +calc_translate (int dst_size, + int src_size, + pixman_fixed_t scale, + pixman_bool_t low_align, + pixman_bool_t bilinear) +{ + pixman_fixed_t ref_src, ref_dst, scaled_dst; + + if (low_align) + { + ref_src = bilinear ? pixman_fixed_1 / 2 : pixman_fixed_e; + ref_dst = pixman_fixed_1 / 2; + } + else + { + ref_src = pixman_int_to_fixed (src_size) - + bilinear * pixman_fixed_1 / 2; + ref_dst = pixman_int_to_fixed (dst_size) - pixman_fixed_1 / 2; + } + + scaled_dst = ((uint64_t) ref_dst * scale + pixman_fixed_1 / 2) / + pixman_fixed_1; + + /* We need the translation to be set such that when ref_dst is fed through + * the transformation matrix, we get ref_src as the result. + */ + return ref_src - scaled_dst; +} + +static pixman_fixed_t +random_offset (void) +{ + pixman_fixed_t offset = 0; + + /* Ensure we test the exact case quite a lot */ + if (prng_rand_n (2)) + return offset; + + /* What happens when we are close to the edge of the first + * interpolation step? + */ + if (prng_rand_n (2)) + offset += (pixman_fixed_1 >> BILINEAR_INTERPOLATION_BITS) - 16; + + /* Try fine-grained variations */ + offset += prng_rand_n (32); + + /* Test in both directions */ + if (prng_rand_n (2)) + offset = -offset; + + return offset; +} + +static void +check_transform (pixman_image_t *dst_img, + pixman_image_t *src_img, + pixman_transform_t *transform, + pixman_bool_t bilinear) +{ + pixman_vector_t v1, v2; + + v1.vector[0] = pixman_fixed_1 / 2; + v1.vector[1] = pixman_fixed_1 / 2; + v1.vector[2] = pixman_fixed_1; + assert (pixman_transform_point (transform, &v1)); + + v2.vector[0] = pixman_int_to_fixed (dst_img->bits.width) - + pixman_fixed_1 / 2; + v2.vector[1] = pixman_int_to_fixed (dst_img->bits.height) - + pixman_fixed_1 / 2; + v2.vector[2] = pixman_fixed_1; + assert (pixman_transform_point (transform, &v2)); + + if (bilinear) + { + assert (v1.vector[0] >= pixman_fixed_1 / 2); + assert (v1.vector[1] >= pixman_fixed_1 / 2); + assert (v2.vector[0] <= pixman_int_to_fixed (src_img->bits.width) - + pixman_fixed_1 / 2); + assert (v2.vector[1] <= pixman_int_to_fixed (src_img->bits.height) - + pixman_fixed_1 / 2); + } + else + { + assert (v1.vector[0] >= pixman_fixed_e); + assert (v1.vector[1] >= pixman_fixed_e); + assert (v2.vector[0] <= pixman_int_to_fixed (src_img->bits.width)); + assert (v2.vector[1] <= pixman_int_to_fixed (src_img->bits.height)); + } +} + +static uint32_t +test_cover (int testnum, int verbose) +{ + pixman_fixed_t x_scale, y_scale; + pixman_bool_t left_align, top_align; + pixman_bool_t bilinear; + pixman_filter_t filter; + pixman_op_t op; + size_t src_fmt_index; + pixman_format_code_t src_fmt, dst_fmt, mask_fmt; + pixman_image_t *src_img, *dst_img, *mask_img; + pixman_transform_t src_transform, mask_transform; + pixman_fixed_t fuzz[4]; + uint32_t crc32; + + /* We allocate one fenced image for each pixel format up-front. This is to + * avoid spending a lot of time on memory management rather than on testing + * Pixman optimisations. We need one per thread because the transformation + * matrices and filtering are properties of the source and mask images. + */ + static pixman_image_t *src_imgs[ARRAY_LENGTH (img_fmt_list)]; + static pixman_image_t *mask_bits_img; + static pixman_bool_t fence_images_created; +#ifdef USE_OPENMP +#pragma omp threadprivate (src_imgs) +#pragma omp threadprivate (mask_bits_img) +#pragma omp threadprivate (fence_images_created) +#endif + + if (!fence_images_created) + { + int i; + + prng_srand (0); + + for (i = 0; i < ARRAY_LENGTH (img_fmt_list); i++) + src_imgs[i] = create_src_image (img_fmt_list[i]); + + mask_bits_img = create_src_image (PIXMAN_a8); + + fence_images_created = TRUE; + } + + prng_srand (testnum); + + x_scale = random_scale_factor (); + y_scale = random_scale_factor (); + left_align = prng_rand_n (2); + top_align = prng_rand_n (2); + bilinear = prng_rand_n (2); + filter = bilinear ? PIXMAN_FILTER_BILINEAR : PIXMAN_FILTER_NEAREST; + + op = op_list[prng_rand_n (ARRAY_LENGTH (op_list))]; + + dst_fmt = img_fmt_list[prng_rand_n (ARRAY_LENGTH (img_fmt_list))]; + dst_img = pixman_image_create_bits (dst_fmt, DST_WIDTH, DST_HEIGHT, + NULL, 0); + prng_randmemset (dst_img->bits.bits, + dst_img->bits.rowstride * DST_HEIGHT * sizeof (uint32_t), + 0); + image_endian_swap (dst_img); + + src_fmt_index = prng_rand_n (ARRAY_LENGTH (img_fmt_list)); + src_fmt = img_fmt_list[src_fmt_index]; + src_img = src_imgs[src_fmt_index]; + pixman_image_set_filter (src_img, filter, NULL, 0); + pixman_transform_init_scale (&src_transform, x_scale, y_scale); + src_transform.matrix[0][2] = calc_translate (dst_img->bits.width, + src_img->bits.width, + x_scale, left_align, bilinear); + src_transform.matrix[1][2] = calc_translate (dst_img->bits.height, + src_img->bits.height, + y_scale, top_align, bilinear); + + if (prng_rand_n (2)) + { + /* No mask */ + mask_fmt = PIXMAN_null; + mask_img = NULL; + } + else if (prng_rand_n (2)) + { + /* a8 bitmap mask */ + mask_fmt = PIXMAN_a8; + mask_img = mask_bits_img; + pixman_image_set_filter (mask_img, filter, NULL, 0); + pixman_transform_init_scale (&mask_transform, x_scale, y_scale); + mask_transform.matrix[0][2] = calc_translate (dst_img->bits.width, + mask_img->bits.width, + x_scale, left_align, + bilinear); + mask_transform.matrix[1][2] = calc_translate (dst_img->bits.height, + mask_img->bits.height, + y_scale, top_align, + bilinear); + } + else + { + /* Solid mask */ + pixman_color_t color; + memset (&color, 0xAA, sizeof color); + mask_fmt = PIXMAN_solid; + mask_img = pixman_image_create_solid_fill (&color); + } + + if (!exact) + { + int i = 0; + + while (i < 4) + fuzz[i++] = random_offset (); + + src_transform.matrix[0][2] += fuzz[0]; + src_transform.matrix[1][2] += fuzz[1]; + mask_transform.matrix[0][2] += fuzz[2]; + mask_transform.matrix[1][2] += fuzz[3]; + } + + pixman_image_set_transform (src_img, &src_transform); + if (mask_fmt == PIXMAN_a8) + pixman_image_set_transform (mask_img, &mask_transform); + + if (verbose) + { + printf ("op=%s\n", operator_name (op)); + printf ("src_fmt=%s, dst_fmt=%s, mask_fmt=%s\n", + format_name (src_fmt), format_name (dst_fmt), + format_name (mask_fmt)); + printf ("x_scale=0x%08X, y_scale=0x%08X, align %s/%s, %s\n", + x_scale, y_scale, + left_align ? "left" : "right", top_align ? "top" : "bottom", + bilinear ? "bilinear" : "nearest"); + + if (!exact) + { + int i = 0; + + printf ("fuzz factors"); + while (i < 4) + printf (" %d", fuzz[i++]); + printf ("\n"); + } + } + + if (exact) + { + check_transform (dst_img, src_img, &src_transform, bilinear); + if (mask_fmt == PIXMAN_a8) + check_transform (dst_img, mask_img, &mask_transform, bilinear); + } + + pixman_image_composite (op, src_img, mask_img, dst_img, + 0, 0, 0, 0, 0, 0, + dst_img->bits.width, dst_img->bits.height); + + if (verbose) + print_image (dst_img); + + crc32 = compute_crc32_for_image (0, dst_img); + + pixman_image_unref (dst_img); + if (mask_fmt == PIXMAN_solid) + pixman_image_unref (mask_img); + + return crc32; +} + +#if BILINEAR_INTERPOLATION_BITS == 7 +#define CHECKSUM_FUZZ 0x6B56F607 +#define CHECKSUM_EXACT 0xA669F4A3 +#elif BILINEAR_INTERPOLATION_BITS == 4 +#define CHECKSUM_FUZZ 0x83119ED0 +#define CHECKSUM_EXACT 0x0D3382CD +#else +#define CHECKSUM_FUZZ 0x00000000 +#define CHECKSUM_EXACT 0x00000000 +#endif + +int +main (int argc, const char *argv[]) +{ + unsigned long page_size; + + page_size = fence_get_page_size (); + if (page_size == 0 || page_size > 16 * 1024) + return 77; /* automake SKIP */ + + exact = getenv ("EXACT") != NULL; + if (exact) + printf ("Doing plots that are exactly aligned to boundaries\n"); + + return fuzzer_test_main ("cover", 2000000, + exact ? CHECKSUM_EXACT : CHECKSUM_FUZZ, + test_cover, argc, argv); +} diff --git a/lib/pixman/test/fence-image-self-test.c b/lib/pixman/test/fence-image-self-test.c new file mode 100644 index 000000000..c80b3cf31 --- /dev/null +++ b/lib/pixman/test/fence-image-self-test.c @@ -0,0 +1,239 @@ +/* + * Copyright © 2015 Raspberry Pi Foundation + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. The copyright holders make no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include "utils.h" + + +#if FENCE_MALLOC_ACTIVE && defined (HAVE_SIGACTION) + +#include +#include +#include +#include +#include +#include +#include +#include + +pixman_bool_t verbose; + +static void +segv_handler (int sig, siginfo_t *si, void *unused) +{ + _exit (EXIT_SUCCESS); +} + +static void +die (const char *msg, int err) +{ + if (err) + perror (msg); + else + fprintf (stderr, "%s\n", msg); + + abort (); +} + +static void +prinfo (const char *fmt, ...) +{ + va_list ap; + + if (!verbose) + return; + + va_start (ap, fmt); + vfprintf (stderr, fmt, ap); + va_end (ap); +} + +static void +do_expect_signal (void (*fn)(void *), void *data) +{ + struct sigaction sa; + + sa.sa_flags = SA_SIGINFO; + sigemptyset (&sa.sa_mask); + sa.sa_sigaction = segv_handler; + if (sigaction (SIGSEGV, &sa, NULL) == -1) + die ("sigaction failed", errno); + if (sigaction (SIGBUS, &sa, NULL) == -1) + die ("sigaction failed", errno); + + (*fn)(data); + + _exit (EXIT_FAILURE); +} + +/* Check that calling fn(data) causes a segmentation fault. + * + * You cannot portably return from a SIGSEGV handler in any way, + * so we fork, and do the test in the child process. Child's + * exit status will reflect the result. Its SEGV handler causes it + * to exit with success, and return failure otherwise. + */ +static pixman_bool_t +expect_signal (void (*fn)(void *), void *data) +{ + pid_t pid, wp; + int status; + + pid = fork (); + if (pid == -1) + die ("fork failed", errno); + + if (pid == 0) + do_expect_signal (fn, data); /* never returns */ + + wp = waitpid (pid, &status, 0); + if (wp != pid) + die ("waitpid did not work", wp == -1 ? errno : 0); + + if (WIFEXITED (status) && WEXITSTATUS (status) == EXIT_SUCCESS) + return TRUE; + + return FALSE; +} + +static void +read_u8 (void *data) +{ + volatile uint8_t *p = data; + + *p; +} + +static pixman_bool_t +test_read_fault (uint8_t *p, int offset) +{ + prinfo ("*(uint8_t *)(%p + %d)", p, offset); + + if (expect_signal (read_u8, p + offset)) + { + prinfo ("\tsignal OK\n"); + + return TRUE; + } + + prinfo ("\tFAILED\n"); + + return FALSE; +} + +static void +test_read_ok (uint8_t *p, int offset) +{ + prinfo ("*(uint8_t *)(%p + %d)", p, offset); + + /* If fails, SEGV. */ + read_u8 (p + offset); + + prinfo ("\tOK\n"); +} + +static pixman_bool_t +test_read_faults (pixman_image_t *image) +{ + pixman_bool_t ok = TRUE; + pixman_format_code_t format = pixman_image_get_format (image); + int width = pixman_image_get_width (image); + int height = pixman_image_get_height (image); + int stride = pixman_image_get_stride (image); + uint8_t *p = (void *)pixman_image_get_data (image); + int row_bytes = width * PIXMAN_FORMAT_BPP (format) / 8; + + prinfo ("%s %dx%d, row %d B, stride %d B:\n", + format_name (format), width, height, row_bytes, stride); + + assert (height > 3); + + test_read_ok (p, 0); + test_read_ok (p, row_bytes - 1); + test_read_ok (p, stride); + test_read_ok (p, stride + row_bytes - 1); + test_read_ok (p, 2 * stride); + test_read_ok (p, 2 * stride + row_bytes - 1); + test_read_ok (p, 3 * stride); + test_read_ok (p, (height - 1) * stride + row_bytes - 1); + + ok &= test_read_fault (p, -1); + ok &= test_read_fault (p, row_bytes); + ok &= test_read_fault (p, stride - 1); + ok &= test_read_fault (p, stride + row_bytes); + ok &= test_read_fault (p, 2 * stride - 1); + ok &= test_read_fault (p, 2 * stride + row_bytes); + ok &= test_read_fault (p, 3 * stride - 1); + ok &= test_read_fault (p, height * stride); + + return ok; +} + +static pixman_bool_t +test_image_faults (pixman_format_code_t format, int min_width, int height) +{ + pixman_bool_t ok; + pixman_image_t *image; + + image = fence_image_create_bits (format, min_width, height, TRUE); + ok = test_read_faults (image); + pixman_image_unref (image); + + return ok; +} + +int +main (int argc, char **argv) +{ + pixman_bool_t ok = TRUE; + + if (getenv ("VERBOSE") != NULL) + verbose = TRUE; + + ok &= test_image_faults (PIXMAN_a8r8g8b8, 7, 5); + ok &= test_image_faults (PIXMAN_r8g8b8, 7, 5); + ok &= test_image_faults (PIXMAN_r5g6b5, 7, 5); + ok &= test_image_faults (PIXMAN_a8, 7, 5); + ok &= test_image_faults (PIXMAN_a4, 7, 5); + ok &= test_image_faults (PIXMAN_a1, 7, 5); + + if (ok) + return EXIT_SUCCESS; + + return EXIT_FAILURE; +} + +#else /* FENCE_MALLOC_ACTIVE */ + +int +main (int argc, char **argv) +{ + /* Automake return code for test SKIP. */ + return 77; +} + +#endif /* FENCE_MALLOC_ACTIVE */ diff --git a/lib/pixman/test/lowlevel-blt-bench.c b/lib/pixman/test/lowlevel-blt-bench.c index 1049e21e7..28ff66991 100644 --- a/lib/pixman/test/lowlevel-blt-bench.c +++ b/lib/pixman/test/lowlevel-blt-bench.c @@ -55,7 +55,7 @@ uint32_t *dst; uint32_t *src; uint32_t *mask; -double bandwidth = 0; +double bandwidth = 0.0; double bench_memcpy () @@ -90,6 +90,7 @@ bench_memcpy () static pixman_bool_t use_scaling = FALSE; static pixman_filter_t filter = PIXMAN_FILTER_NEAREST; +static pixman_bool_t use_csv_output = FALSE; /* nearly 1x scale factor */ static pixman_transform_t m = @@ -165,7 +166,7 @@ call_func (pixman_composite_func_t func, func (0, &info); } -void +double noinline bench_L (pixman_op_t op, pixman_image_t * src_img, @@ -204,9 +205,11 @@ bench_L (pixman_op_t op, call_func (func, op, src_img, mask_img, dst_img, x, 0, x, 0, 63 - x, 0, width, lines_count); } qx = q; + + return (double)n * lines_count * width; } -void +double noinline bench_M (pixman_op_t op, pixman_image_t * src_img, @@ -224,6 +227,8 @@ bench_M (pixman_op_t op, x = 0; call_func (func, op, src_img, mask_img, dst_img, x, 0, x, 0, 1, 0, WIDTH - 64, HEIGHT); } + + return (double)n * (WIDTH - 64) * HEIGHT; } double @@ -366,15 +371,24 @@ bench_RT (pixman_op_t op, return pix_cnt; } +static double +Mpx_per_sec (double pix_cnt, double t1, double t2, double t3) +{ + double overhead = t2 - t1; + double testtime = t3 - t2; + + return pix_cnt / (testtime - overhead) / 1e6; +} + void -bench_composite (char * testname, - int src_fmt, - int src_flags, - int op, - int mask_fmt, - int mask_flags, - int dst_fmt, - double npix) +bench_composite (const char *testname, + int src_fmt, + int src_flags, + int op, + int mask_fmt, + int mask_flags, + int dst_fmt, + double npix) { pixman_image_t * src_img; pixman_image_t * dst_img; @@ -461,9 +475,9 @@ bench_composite (char * testname, dst, XWIDTH * 4); - - printf ("%24s %c", testname, func != pixman_image_composite_wrapper ? - '-' : '='); + if (!use_csv_output) + printf ("%24s %c", testname, func != pixman_image_composite_wrapper ? + '-' : '='); memcpy (dst, src, BUFSIZE); memcpy (src, dst, BUFSIZE); @@ -476,13 +490,15 @@ bench_composite (char * testname, n = 1 + npix / (l1test_width * 8); t1 = gettime (); #if EXCLUDE_OVERHEAD - bench_L (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty, l1test_width, 1); + pix_cnt = bench_L (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty, l1test_width, 1); #endif t2 = gettime (); - bench_L (op, src_img, mask_img, dst_img, n, func, l1test_width, 1); + pix_cnt = bench_L (op, src_img, mask_img, dst_img, n, func, l1test_width, 1); t3 = gettime (); - printf (" L1:%7.2f", (double)n * l1test_width * 1 / - ((t3 - t2) - (t2 - t1)) / 1000000.); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" L1:%7.2f", Mpx_per_sec (pix_cnt, t1, t2, t3)); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -495,13 +511,15 @@ bench_composite (char * testname, n = 1 + npix / (l1test_width * nlines); t1 = gettime (); #if EXCLUDE_OVERHEAD - bench_L (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty, l1test_width, nlines); + pix_cnt = bench_L (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty, l1test_width, nlines); #endif t2 = gettime (); - bench_L (op, src_img, mask_img, dst_img, n, func, l1test_width, nlines); + pix_cnt = bench_L (op, src_img, mask_img, dst_img, n, func, l1test_width, nlines); t3 = gettime (); - printf (" L2:%7.2f", (double)n * l1test_width * nlines / - ((t3 - t2) - (t2 - t1)) / 1000000.); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" L2:%7.2f", Mpx_per_sec (pix_cnt, t1, t2, t3)); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -510,14 +528,16 @@ bench_composite (char * testname, n = 1 + npix / (WIDTH * HEIGHT); t1 = gettime (); #if EXCLUDE_OVERHEAD - bench_M (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty); + pix_cnt = bench_M (op, src_img, mask_img, dst_img, n, pixman_image_composite_empty); #endif t2 = gettime (); - bench_M (op, src_img, mask_img, dst_img, n, func); + pix_cnt = bench_M (op, src_img, mask_img, dst_img, n, func); t3 = gettime (); - printf (" M:%6.2f (%6.2f%%)", - ((double)n * (WIDTH - 64) * HEIGHT / ((t3 - t2) - (t2 - t1))) / 1000000., - ((double)n * (WIDTH - 64) * HEIGHT / ((t3 - t2) - (t2 - t1)) * bytes_per_pix) * (100.0 / bandwidth) ); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" M:%6.2f (%6.2f%%)", Mpx_per_sec (pix_cnt, t1, t2, t3), + (pix_cnt / ((t3 - t2) - (t2 - t1)) * bytes_per_pix) * (100.0 / bandwidth) ); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -531,7 +551,10 @@ bench_composite (char * testname, t2 = gettime (); pix_cnt = bench_HT (op, src_img, mask_img, dst_img, n, func); t3 = gettime (); - printf (" HT:%6.2f", (double)pix_cnt / ((t3 - t2) - (t2 - t1)) / 1000000.); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" HT:%6.2f", Mpx_per_sec (pix_cnt, t1, t2, t3)); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -545,7 +568,10 @@ bench_composite (char * testname, t2 = gettime (); pix_cnt = bench_VT (op, src_img, mask_img, dst_img, n, func); t3 = gettime (); - printf (" VT:%6.2f", (double)pix_cnt / ((t3 - t2) - (t2 - t1)) / 1000000.); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" VT:%6.2f", Mpx_per_sec (pix_cnt, t1, t2, t3)); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -559,7 +585,10 @@ bench_composite (char * testname, t2 = gettime (); pix_cnt = bench_R (op, src_img, mask_img, dst_img, n, func, WIDTH, HEIGHT); t3 = gettime (); - printf (" R:%6.2f", (double)pix_cnt / ((t3 - t2) - (t2 - t1)) / 1000000.); + if (use_csv_output) + printf ("%g,", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" R:%6.2f", Mpx_per_sec (pix_cnt, t1, t2, t3)); fflush (stdout); memcpy (dst, src, BUFSIZE); @@ -573,7 +602,10 @@ bench_composite (char * testname, t2 = gettime (); pix_cnt = bench_RT (op, src_img, mask_img, dst_img, n, func, WIDTH, HEIGHT); t3 = gettime (); - printf (" RT:%6.2f (%4.0fKops/s)\n", (double)pix_cnt / ((t3 - t2) - (t2 - t1)) / 1000000., (double) n / ((t3 - t2) * 1000)); + if (use_csv_output) + printf ("%g\n", Mpx_per_sec (pix_cnt, t1, t2, t3)); + else + printf (" RT:%6.2f (%4.0fKops/s)\n", Mpx_per_sec (pix_cnt, t1, t2, t3), (double) n / ((t3 - t2) * 1000)); if (mask_img) { pixman_image_unref (mask_img); @@ -587,17 +619,20 @@ bench_composite (char * testname, #define PIXMAN_OP_OUT_REV (PIXMAN_OP_OUT_REVERSE) -struct +struct test_entry { - char *testname; - int src_fmt; - int src_flags; - int op; - int mask_fmt; - int mask_flags; - int dst_fmt; -} -tests_tbl[] = + const char *testname; + int src_fmt; + int src_flags; + int op; + int mask_fmt; + int mask_flags; + int dst_fmt; +}; + +typedef struct test_entry test_entry_t; + +static const test_entry_t tests_tbl[] = { { "add_8_8_8", PIXMAN_a8, 0, PIXMAN_OP_ADD, PIXMAN_a8, 0, PIXMAN_a8 }, { "add_n_8_8", PIXMAN_a8r8g8b8, 1, PIXMAN_OP_ADD, PIXMAN_a8, 0, PIXMAN_a8 }, @@ -713,51 +748,292 @@ tests_tbl[] = { "outrev_n_8888_1555_ca", PIXMAN_a8r8g8b8, 1, PIXMAN_OP_OUT_REV, PIXMAN_a8r8g8b8, 2, PIXMAN_a1r5g5b5 }, { "outrev_n_8888_x888_ca", PIXMAN_a8r8g8b8, 1, PIXMAN_OP_OUT_REV, PIXMAN_a8r8g8b8, 2, PIXMAN_x8r8g8b8 }, { "outrev_n_8888_8888_ca", PIXMAN_a8r8g8b8, 1, PIXMAN_OP_OUT_REV, PIXMAN_a8r8g8b8, 2, PIXMAN_a8r8g8b8 }, - { "over_reverse_n_8888", PIXMAN_a8r8g8b8, 0, PIXMAN_OP_OVER_REVERSE, PIXMAN_null, 0, PIXMAN_a8r8g8b8 }, + { "over_reverse_n_8888", PIXMAN_a8r8g8b8, 1, PIXMAN_OP_OVER_REVERSE, PIXMAN_null, 0, PIXMAN_a8r8g8b8 }, + { "in_reverse_8888_8888", PIXMAN_a8r8g8b8, 0, PIXMAN_OP_IN_REVERSE, PIXMAN_null, 0, PIXMAN_a8r8g8b8 }, { "pixbuf", PIXMAN_x8b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_a8b8g8r8, 0, PIXMAN_a8r8g8b8 }, { "rpixbuf", PIXMAN_x8b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_a8b8g8r8, 0, PIXMAN_a8b8g8r8 }, }; -int -main (int argc, char *argv[]) +static const test_entry_t special_patterns[] = +{ + { "add_n_2x10", PIXMAN_a2r10g10b10, 1, PIXMAN_OP_ADD, PIXMAN_null, 0, PIXMAN_x2r10g10b10 }, + { "add_n_2a10", PIXMAN_a2r10g10b10, 1, PIXMAN_OP_ADD, PIXMAN_null, 0, PIXMAN_a2r10g10b10 }, + { "src_n_2x10", PIXMAN_a2r10g10b10, 1, PIXMAN_OP_SRC, PIXMAN_null, 0, PIXMAN_x2r10g10b10 }, + { "src_n_2a10", PIXMAN_a2r10g10b10, 1, PIXMAN_OP_SRC, PIXMAN_null, 0, PIXMAN_a2r10g10b10 }, + { "src_0888_8888_rev", PIXMAN_b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_null, 0, PIXMAN_x8r8g8b8 }, + { "src_0888_0565_rev", PIXMAN_b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_null, 0, PIXMAN_r5g6b5 }, + { "src_n_8", PIXMAN_a8, 1, PIXMAN_OP_SRC, PIXMAN_null, 0, PIXMAN_a8 }, + { "pixbuf", PIXMAN_x8b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_a8b8g8r8, 0, PIXMAN_a8r8g8b8 }, + { "rpixbuf", PIXMAN_x8b8g8r8, 0, PIXMAN_OP_SRC, PIXMAN_a8b8g8r8, 0, PIXMAN_a8b8g8r8 }, +}; + +/* Returns the sub-string's end pointer in string. */ +static const char * +copy_sub_string (char *buf, + const char *string, + const char *scan_from, + const char *end) { - double x; + const char *delim; + size_t n; + + delim = strchr (scan_from, '_'); + if (!delim) + delim = end; + + n = delim - string; + strncpy(buf, string, n); + buf[n] = '\0'; + + return delim; +} + +static pixman_op_t +parse_longest_operator (char *buf, const char **strp, const char *end) +{ + const char *p = *strp; + const char *sub_end; + const char *best_end = p; + pixman_op_t best_op = PIXMAN_OP_NONE; + pixman_op_t op; + + while (p < end) + { + sub_end = copy_sub_string (buf, *strp, p, end); + op = operator_from_string (buf); + p = sub_end + 1; + + if (op != PIXMAN_OP_NONE) + { + best_end = p; + best_op = op; + } + } + + *strp = best_end; + return best_op; +} + +static pixman_format_code_t +parse_format (char *buf, const char **p, const char *end) +{ + pixman_format_code_t format; + const char *delim; + + if (*p >= end) + return PIXMAN_null; + + delim = copy_sub_string (buf, *p, *p, end); + format = format_from_string (buf); + + if (format != PIXMAN_null) + *p = delim + 1; + + return format; +} + +static int +parse_test_pattern (test_entry_t *test, const char *pattern) +{ + const char *p = pattern; + const char *end = pattern + strlen (pattern); + char buf[1024]; + pixman_format_code_t format[3]; int i; - const char *pattern = NULL; - for (i = 1; i < argc; i++) + + if (strlen (pattern) > sizeof (buf) - 1) + return -1; + + /* Special cases that the parser cannot produce. */ + for (i = 0; i < ARRAY_LENGTH (special_patterns); i++) { - if (argv[i][0] == '-') - { - if (strchr (argv[i] + 1, 'b')) - { - use_scaling = TRUE; - filter = PIXMAN_FILTER_BILINEAR; - } - else if (strchr (argv[i] + 1, 'n')) - { - use_scaling = TRUE; - filter = PIXMAN_FILTER_NEAREST; - } - } - else - { - pattern = argv[i]; - } + if (strcmp (pattern, special_patterns[i].testname) == 0) + { + *test = special_patterns[i]; + return 0; + } } - if (!pattern) + test->testname = pattern; + + /* Extract operator, may contain delimiters, + * so take the longest string that matches. + */ + test->op = parse_longest_operator (buf, &p, end); + if (test->op == PIXMAN_OP_NONE) + return -1; + + /* extract up to three pixel formats */ + format[0] = parse_format (buf, &p, end); + format[1] = parse_format (buf, &p, end); + format[2] = parse_format (buf, &p, end); + + if (format[0] == PIXMAN_null || format[1] == PIXMAN_null) + return -1; + + /* recognize CA flag */ + test->mask_flags = 0; + if (p < end) { - printf ("Usage: lowlevel-blt-bench [-b] [-n] pattern\n"); - printf (" -n : benchmark nearest scaling\n"); - printf (" -b : benchmark bilinear scaling\n"); - return 1; + if (strcmp (p, "ca") == 0) + test->mask_flags |= CA_FLAG; + else + return -1; /* trailing garbage */ } - src = aligned_malloc (4096, BUFSIZE * 3); - memset (src, 0xCC, BUFSIZE * 3); - dst = src + (BUFSIZE / 4); - mask = dst + (BUFSIZE / 4); + test->src_fmt = format[0]; + if (format[2] == PIXMAN_null) + { + test->mask_fmt = PIXMAN_null; + test->dst_fmt = format[1]; + } + else + { + test->mask_fmt = format[1]; + test->dst_fmt = format[2]; + } + + test->src_flags = 0; + if (test->src_fmt == PIXMAN_solid) + { + test->src_fmt = PIXMAN_a8r8g8b8; + test->src_flags |= SOLID_FLAG; + } + + if (test->mask_fmt == PIXMAN_solid) + { + if (test->mask_flags & CA_FLAG) + test->mask_fmt = PIXMAN_a8r8g8b8; + else + test->mask_fmt = PIXMAN_a8; + + test->mask_flags |= SOLID_FLAG; + } + + return 0; +} +static int +check_int (int got, int expected, const char *name, const char *field) +{ + if (got == expected) + return 0; + + printf ("%s: %s failure: expected %d, got %d.\n", + name, field, expected, got); + + return 1; +} + +static int +check_format (int got, int expected, const char *name, const char *field) +{ + if (got == expected) + return 0; + + printf ("%s: %s failure: expected %s (%#x), got %s (%#x).\n", + name, field, + format_name (expected), expected, + format_name (got), got); + + return 1; +} + +static void +parser_self_test (void) +{ + const test_entry_t *ent; + test_entry_t test; + int fails = 0; + int i; + + for (i = 0; i < ARRAY_LENGTH (tests_tbl); i++) + { + ent = &tests_tbl[i]; + + if (parse_test_pattern (&test, ent->testname) < 0) + { + printf ("parsing failed for '%s'\n", ent->testname); + fails++; + continue; + } + + fails += check_format (test.src_fmt, ent->src_fmt, + ent->testname, "src_fmt"); + fails += check_format (test.mask_fmt, ent->mask_fmt, + ent->testname, "mask_fmt"); + fails += check_format (test.dst_fmt, ent->dst_fmt, + ent->testname, "dst_fmt"); + fails += check_int (test.src_flags, ent->src_flags, + ent->testname, "src_flags"); + fails += check_int (test.mask_flags, ent->mask_flags, + ent->testname, "mask_flags"); + fails += check_int (test.op, ent->op, ent->testname, "op"); + } + + if (fails) + { + printf ("Parser self-test failed.\n"); + exit (EXIT_FAILURE); + } + + if (!use_csv_output) + printf ("Parser self-test complete.\n"); +} + +static void +print_test_details (const test_entry_t *test) +{ + printf ("%s: %s, src %s%s, mask %s%s%s, dst %s\n", + test->testname, + operator_name (test->op), + format_name (test->src_fmt), + test->src_flags & SOLID_FLAG ? " solid" : "", + format_name (test->mask_fmt), + test->mask_flags & SOLID_FLAG ? " solid" : "", + test->mask_flags & CA_FLAG ? " CA" : "", + format_name (test->dst_fmt)); +} + +static void +run_one_test (const char *pattern, double bandwidth_, pixman_bool_t prdetails) +{ + test_entry_t test; + + if (parse_test_pattern (&test, pattern) < 0) + { + printf ("Error: Could not parse the test pattern '%s'.\n", pattern); + return; + } + + if (prdetails) + { + print_test_details (&test); + printf ("---\n"); + } + + bench_composite (pattern, + test.src_fmt, + test.src_flags, + test.op, + test.mask_fmt, + test.mask_flags, + test.dst_fmt, + bandwidth_ / 8); +} + +static void +run_default_tests (double bandwidth_) +{ + int i; + + for (i = 0; i < ARRAY_LENGTH (tests_tbl); i++) + run_one_test (tests_tbl[i].testname, bandwidth_, FALSE); +} + +static void +print_explanation (void) +{ printf ("Benchmark for a set of most commonly used functions\n"); printf ("---\n"); printf ("All results are presented in millions of pixels per second\n"); @@ -785,9 +1061,14 @@ main (int argc, char *argv[]) printf ("RT - as R, but %dx%d average sized rectangles are copied\n", TINYWIDTH, TINYWIDTH); printf ("---\n"); - bandwidth = x = bench_memcpy (); +} + +static void +print_speed_scaling (double bw) +{ printf ("reference memcpy speed = %.1fMB/s (%.1fMP/s for 32bpp fills)\n", - x / 1000000., x / 4000000); + bw / 1000000., bw / 4000000); + if (use_scaling) { printf ("---\n"); @@ -798,23 +1079,85 @@ main (int argc, char *argv[]) else printf ("UNKNOWN scaling\n"); } + printf ("---\n"); +} - for (i = 0; i < ARRAY_LENGTH (tests_tbl); i++) +static void +usage (const char *progname) +{ + printf ("Usage: %s [-b] [-n] [-c] [-m M] pattern\n", progname); + printf (" -n : benchmark nearest scaling\n"); + printf (" -b : benchmark bilinear scaling\n"); + printf (" -c : print output as CSV data\n"); + printf (" -m M : set reference memcpy speed to M MB/s instead of measuring it\n"); +} + +int +main (int argc, char *argv[]) +{ + int i; + const char *pattern = NULL; + + for (i = 1; i < argc; i++) { - if (strcmp (pattern, "all") == 0 || strcmp (tests_tbl[i].testname, pattern) == 0) + if (argv[i][0] == '-') { - bench_composite (tests_tbl[i].testname, - tests_tbl[i].src_fmt, - tests_tbl[i].src_flags, - tests_tbl[i].op, - tests_tbl[i].mask_fmt, - tests_tbl[i].mask_flags, - tests_tbl[i].dst_fmt, - bandwidth/8); + if (strchr (argv[i] + 1, 'b')) + { + use_scaling = TRUE; + filter = PIXMAN_FILTER_BILINEAR; + } + else if (strchr (argv[i] + 1, 'n')) + { + use_scaling = TRUE; + filter = PIXMAN_FILTER_NEAREST; + } + + if (strchr (argv[i] + 1, 'c')) + use_csv_output = TRUE; + + if (strcmp (argv[i], "-m") == 0 && i + 1 < argc) + bandwidth = atof (argv[++i]) * 1e6; } + else + { + if (pattern) + { + pattern = NULL; + printf ("Error: extra arguments given.\n"); + break; + } + pattern = argv[i]; + } + } + + if (!pattern) + { + usage (argv[0]); + return 1; } + parser_self_test (); + + src = aligned_malloc (4096, BUFSIZE * 3); + memset (src, 0xCC, BUFSIZE * 3); + dst = src + (BUFSIZE / 4); + mask = dst + (BUFSIZE / 4); + + if (!use_csv_output) + print_explanation (); + + if (bandwidth < 1.0) + bandwidth = bench_memcpy (); + if (!use_csv_output) + print_speed_scaling (bandwidth); + + if (strcmp (pattern, "all") == 0) + run_default_tests (bandwidth); + else + run_one_test (pattern, bandwidth, !use_csv_output); + free (src); return 0; } diff --git a/lib/pixman/test/pixel-test.c b/lib/pixman/test/pixel-test.c index 8c525d202..7dc0eff3f 100644 --- a/lib/pixman/test/pixel-test.c +++ b/lib/pixman/test/pixel-test.c @@ -33,68 +33,2764 @@ struct pixel_combination_t pixman_op_t op; pixman_format_code_t src_format; uint32_t src_pixel; + pixman_format_code_t mask_format; + uint32_t mask_pixel; pixman_format_code_t dest_format; uint32_t dest_pixel; }; static const pixel_combination_t regressions[] = { + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffc3ff, + PIXMAN_a8, 0x7b, + PIXMAN_a8r8g8b8, 0xff00c300, + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb5, + PIXMAN_a4r4g4b4, 0xe3ff, + PIXMAN_a2r2g2b2, 0x2e + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa6, + PIXMAN_a8r8g8b8, 0x2b00ff00, + PIXMAN_a4r4g4b4, 0x7e + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0x27000013, + PIXMAN_a2r2g2b2, 0x80, + PIXMAN_a4r4g4b4, 0x9d + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xe6f7, + PIXMAN_a2r2g2b2, 0xad, + PIXMAN_a4r4g4b4, 0x71 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0xff4f70ff, + PIXMAN_r5g6b5, 0xb828, + PIXMAN_a8r8g8b8, 0xcac400 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa9, + PIXMAN_a4r4g4b4, 0x41c2, + PIXMAN_a8r8g8b8, 0xffff2b + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x89, + PIXMAN_a8r8g8b8, 0x977cff61, + PIXMAN_a4r4g4b4, 0x36 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x81, + PIXMAN_r5g6b5, 0x6f9e, + PIXMAN_a4r4g4b4, 0x1eb + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb5, + PIXMAN_a4r4g4b4, 0xe247, + PIXMAN_a8r8g8b8, 0xffbaff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x97, + PIXMAN_a2r2g2b2, 0x9d, + PIXMAN_a2r2g2b2, 0x21 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb4, + PIXMAN_a2r2g2b2, 0x90, + PIXMAN_a8r8g8b8, 0xc0fd5c + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0xdf00ff70, + PIXMAN_a8r8g8b8, 0x2597ff27, + PIXMAN_a4r4g4b4, 0xf3 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb7, + PIXMAN_r3g3b2, 0xb1, + PIXMAN_a8r8g8b8, 0x9f4bcc + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xf39e, + PIXMAN_r5g6b5, 0x34, + PIXMAN_a8r8g8b8, 0xf6ae00 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0x3aff1dff, + PIXMAN_a2r2g2b2, 0x64, + PIXMAN_a8r8g8b8, 0x94ffb4 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa4, + PIXMAN_a2r2g2b2, 0x8a, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa5, + PIXMAN_a4r4g4b4, 0x1a, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb4, + PIXMAN_a2r2g2b2, 0xca, + PIXMAN_a4r4g4b4, 0x7b + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xbd, + PIXMAN_a4r4g4b4, 0xff37, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x96, + PIXMAN_a2r2g2b2, 0xbb, + PIXMAN_a8r8g8b8, 0x96ffff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x89, + PIXMAN_r3g3b2, 0x92, + PIXMAN_a4r4g4b4, 0xa8c + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xa95b, + PIXMAN_a2r2g2b2, 0x68, + PIXMAN_a8r8g8b8, 0x38ff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x90, + PIXMAN_a8r8g8b8, 0x53bd00ef, + PIXMAN_a8r8g8b8, 0xff0003 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1f5ffff, + PIXMAN_r3g3b2, 0x22, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x10000b6, + PIXMAN_a8r8g8b8, 0x9645, + PIXMAN_r5g6b5, 0x6 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x172ff00, + PIXMAN_a4r4g4b4, 0xff61, + PIXMAN_r3g3b2, 0xc + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x281ffc8, + PIXMAN_r5g6b5, 0x39b8, + PIXMAN_r5g6b5, 0x13 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x100a2ff, + PIXMAN_a4r4g4b4, 0x6500, + PIXMAN_a2r2g2b2, 0x5 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff51, + PIXMAN_r5g6b5, 0x52ff, + PIXMAN_a2r2g2b2, 0x14 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x150d500, + PIXMAN_a8r8g8b8, 0x6200b7ff, + PIXMAN_a8r8g8b8, 0x1f5200 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2a9a700, + PIXMAN_a8r8g8b8, 0xf7003400, + PIXMAN_a8r8g8b8, 0x2200 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x200ffff, + PIXMAN_r5g6b5, 0x81ff, + PIXMAN_r5g6b5, 0x1f + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ff00ff, + PIXMAN_r5g6b5, 0x3f00, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3ff1aa4, + PIXMAN_a4r4g4b4, 0x2200, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x280ff2c, + PIXMAN_r3g3b2, 0xc6, + PIXMAN_a8r8g8b8, 0xfdfd44fe + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x13aff1d, + PIXMAN_a2r2g2b2, 0x4b, + PIXMAN_r5g6b5, 0x12a1 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2ffff88, + PIXMAN_a8r8g8b8, 0xff3a49, + PIXMAN_r5g6b5, 0xf7df + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1009700, + PIXMAN_a2r2g2b2, 0x56, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1aacbff, + PIXMAN_a4r4g4b4, 0x84, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x100b1ff, + PIXMAN_a2r2g2b2, 0xf5, + PIXMAN_a8r8g8b8, 0xfea89cff + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ff0000, + PIXMAN_r5g6b5, 0x6800, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x10064ff, + PIXMAN_r3g3b2, 0x61, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1bb00ff, + PIXMAN_r5g6b5, 0x76b5, + PIXMAN_a4r4g4b4, 0x500 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ffff41, + PIXMAN_r5g6b5, 0x7100, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ff1231, + PIXMAN_a8r8g8b8, 0x381089, + PIXMAN_r5g6b5, 0x38a5 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x16e5c49, + PIXMAN_a8r8g8b8, 0x4dfa3694, + PIXMAN_a8r8g8b8, 0x211c16 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x134ff62, + PIXMAN_a2r2g2b2, 0x14, + PIXMAN_r3g3b2, 0x8 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x300ffeb, + PIXMAN_r3g3b2, 0xc7, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x3ff8bff, + PIXMAN_r3g3b2, 0x3e, + PIXMAN_a8r8g8b8, 0x3008baa + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff00ff, + PIXMAN_a4r4g4b4, 0x3466, + PIXMAN_a4r4g4b4, 0x406 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ddc027, + PIXMAN_a4r4g4b4, 0x7d00, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x2ffff00, + PIXMAN_a8r8g8b8, 0xc92cfb52, + PIXMAN_a4r4g4b4, 0x200 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ff116a, + PIXMAN_a4r4g4b4, 0x6000, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r5g6b5, 0x2f95, + PIXMAN_r5g6b5, 0x795 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffff00, + PIXMAN_a4r4g4b4, 0x354a, + PIXMAN_r5g6b5, 0x3180 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1d7ff00, + PIXMAN_a4r4g4b4, 0xd6ff, + PIXMAN_a8r8g8b8, 0xffff0700 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1bc5db7, + PIXMAN_r5g6b5, 0x944f, + PIXMAN_a4r4g4b4, 0xff05 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x185ffd9, + PIXMAN_a2r2g2b2, 0x9c, + PIXMAN_r5g6b5, 0x3c07 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1fa7f61, + PIXMAN_a8r8g8b8, 0xff31ff00, + PIXMAN_r3g3b2, 0xd2 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1c4ff00, + PIXMAN_r3g3b2, 0xb, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ff00ff, + PIXMAN_a8r8g8b8, 0x3f3caeda, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ff00, + PIXMAN_r5g6b5, 0xff, + PIXMAN_r5g6b5, 0xe0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff68ff, + PIXMAN_a4r4g4b4, 0x8046, + PIXMAN_r5g6b5, 0xec + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x100ff28, + PIXMAN_a8r8g8b8, 0x4c00, + PIXMAN_r5g6b5, 0x260 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_a4r4g4b4, 0xd92a, + PIXMAN_a8r8g8b8, 0x2200 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100289a, + PIXMAN_a8r8g8b8, 0x74ffb8ff, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1baff00, + PIXMAN_r5g6b5, 0x4e9d, + PIXMAN_r5g6b5, 0x3000 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1fcffad, + PIXMAN_r5g6b5, 0x42d7, + PIXMAN_a8r8g8b8, 0x1c6ffe5 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x183ff00, + PIXMAN_r3g3b2, 0x7e, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x2ff0076, + PIXMAN_a8r8g8b8, 0x2a0000, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3d8bbff, + PIXMAN_r5g6b5, 0x6900, + PIXMAN_a8r8g8b8, 0x35b0000 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x14f00ff, + PIXMAN_r5g6b5, 0xd48, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x28c72df, + PIXMAN_a8r8g8b8, 0xff5cff31, + PIXMAN_a4r4g4b4, 0x2 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffffff, + PIXMAN_a8r8g8b8, 0xffad8020, + PIXMAN_r5g6b5, 0x4 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x100ff00, + PIXMAN_a2r2g2b2, 0x76, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1005d00, + PIXMAN_r5g6b5, 0x7b04, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x3cdfc3e, + PIXMAN_a8r8g8b8, 0x69ec21d3, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x200ffff, + PIXMAN_r5g6b5, 0x30ff, + PIXMAN_r5g6b5, 0x60ff + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x532fff4, + PIXMAN_r5g6b5, 0xcb, + PIXMAN_r5g6b5, 0xd9a1 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r3g3b2, 0x5f, + PIXMAN_a2r2g2b2, 0x10 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_a8r8g8b8, 0xffd60052, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ff6491, + PIXMAN_a8r8g8b8, 0x1e53ff00, + PIXMAN_r5g6b5, 0x1862 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_r3g3b2, 0xc7, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x29d0fff, + PIXMAN_a4r4g4b4, 0x25ff, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x141760a, + PIXMAN_a4r4g4b4, 0x7ec2, + PIXMAN_a4r4g4b4, 0x130 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1abedff, + PIXMAN_a8r8g8b8, 0x75520068, + PIXMAN_r3g3b2, 0x87 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x10000ff, + PIXMAN_a8r8g8b8, 0xff00e652, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x16006075, + PIXMAN_r5g6b5, 0xc00, + PIXMAN_a8r8g8b8, 0x27f0900 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x200ff00, + PIXMAN_a8r8g8b8, 0xd1b83f57, + PIXMAN_a4r4g4b4, 0xff75 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x14000c4, + PIXMAN_a4r4g4b4, 0x96, + PIXMAN_a2r2g2b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ff00d1, + PIXMAN_r3g3b2, 0x79, + PIXMAN_a2r2g2b2, 0x0 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff00dc, + PIXMAN_a4r4g4b4, 0xc5ff, + PIXMAN_a2r2g2b2, 0x10 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffb2, + PIXMAN_a8r8g8b8, 0x4cff5700, + PIXMAN_r3g3b2, 0x48 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1327482, + PIXMAN_a8r8g8b8, 0x247ff, + PIXMAN_a8r8g8b8, 0x82 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1d0ff00, + PIXMAN_r3g3b2, 0xc9, + PIXMAN_r5g6b5, 0x240 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x13d35ff, + PIXMAN_a2r2g2b2, 0x6d, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffc6b2, + PIXMAN_a8r8g8b8, 0x5abe8e3c, + PIXMAN_r5g6b5, 0x5a27 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x15700ff, + PIXMAN_r3g3b2, 0xdd, + PIXMAN_a8r8g8b8, 0x55 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff11ff, + PIXMAN_r3g3b2, 0x30, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff00ff, + PIXMAN_a2r2g2b2, 0x6d, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1421d5f, + PIXMAN_a4r4g4b4, 0xff85, + PIXMAN_a8r8g8b8, 0x1420f00 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1d2ffff, + PIXMAN_r5g6b5, 0xfc, + PIXMAN_r5g6b5, 0x1c + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffff42, + PIXMAN_a4r4g4b4, 0x7100, + PIXMAN_a4r4g4b4, 0x771 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x25ae3d4, + PIXMAN_a8r8g8b8, 0x39ffc99a, + PIXMAN_a8r8g8b8, 0x14332f + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff0643, + PIXMAN_a8r8g8b8, 0x4c000000, + PIXMAN_r5g6b5, 0x4802 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1966a00, + PIXMAN_r3g3b2, 0x46, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x387ff59, + PIXMAN_r5g6b5, 0x512c, + PIXMAN_r5g6b5, 0x120 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1f7ffb0, + PIXMAN_r5g6b5, 0x63b8, + PIXMAN_a8r8g8b8, 0x1000089 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x185841c, + PIXMAN_a2r2g2b2, 0x5c, + PIXMAN_a8r8g8b8, 0x8400 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffc3ff, + PIXMAN_a8r8g8b8, 0xff7b, + PIXMAN_a8r8g8b8, 0xff00c300 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff7500, + PIXMAN_a2r2g2b2, 0x47, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1002361, + PIXMAN_a2r2g2b2, 0x7e, + PIXMAN_r5g6b5, 0x64 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x10000b6, + PIXMAN_a8r8g8b8, 0x59004463, + PIXMAN_a4r4g4b4, 0xffa7 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff5a49, + PIXMAN_a8r8g8b8, 0xff3fff2b, + PIXMAN_a8r8g8b8, 0x13f000c + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x3ffecfc, + PIXMAN_r3g3b2, 0x3c, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1630044, + PIXMAN_a2r2g2b2, 0x63, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1d2ff58, + PIXMAN_a8r8g8b8, 0x8f77ff, + PIXMAN_a4r4g4b4, 0x705 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x14dffff, + PIXMAN_a2r2g2b2, 0x9a, + PIXMAN_a8r8g8b8, 0x1a0000 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ff92, + PIXMAN_a4r4g4b4, 0x540c, + PIXMAN_r5g6b5, 0x2a6 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_a4r4g4b4, 0xddd5, + PIXMAN_a4r4g4b4, 0xdd0 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r5g6b5, 0xff8c, + PIXMAN_a4r4g4b4, 0xff0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r3g3b2, 0x66, + PIXMAN_r5g6b5, 0x7d1f + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_a4r4g4b4, 0xff5b, + PIXMAN_a8r8g8b8, 0x5500 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2ed2dff, + PIXMAN_r5g6b5, 0x7ae7, + PIXMAN_r3g3b2, 0xce + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1b13205, + PIXMAN_a8r8g8b8, 0x35ffff00, + PIXMAN_r5g6b5, 0x2040 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1e60dff, + PIXMAN_a4r4g4b4, 0x760f, + PIXMAN_a2r2g2b2, 0x11 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x10000ff, + PIXMAN_a4r4g4b4, 0x3, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_a8r8g8b8, 0x6600, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x30000fa, + PIXMAN_a4r4g4b4, 0x23b7, + PIXMAN_a8r8g8b8, 0x21 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r3g3b2, 0x60, + PIXMAN_r3g3b2, 0x60 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3b31b30, + PIXMAN_r3g3b2, 0x2e, + PIXMAN_a8r8g8b8, 0x3000c20 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x160ffff, + PIXMAN_a4r4g4b4, 0xff42, + PIXMAN_r3g3b2, 0xed + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x172ffff, + PIXMAN_a4r4g4b4, 0x5100, + PIXMAN_r3g3b2, 0x29 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x16300ff, + PIXMAN_a4r4g4b4, 0x5007, + PIXMAN_a8r8g8b8, 0x77 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x2ffff3a, + PIXMAN_a8r8g8b8, 0x26640083, + PIXMAN_a4r4g4b4, 0x220 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x106ff60, + PIXMAN_r5g6b5, 0xdce, + PIXMAN_a8r8g8b8, 0x100ba00 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100e7ff, + PIXMAN_r5g6b5, 0xa00, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2b500f1, + PIXMAN_a4r4g4b4, 0x7339, + PIXMAN_a8r8g8b8, 0x1000091 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff00ff, + PIXMAN_a4r4g4b4, 0xc863, + PIXMAN_r5g6b5, 0x6 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1ffffca, + PIXMAN_a8r8g8b8, 0x8b4cf000, + PIXMAN_r3g3b2, 0xd2 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1fffe00, + PIXMAN_r3g3b2, 0x88, + PIXMAN_r3g3b2, 0x8 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x16f0000, + PIXMAN_a2r2g2b2, 0x59, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x377ff43, + PIXMAN_a4r4g4b4, 0x2a, + PIXMAN_a8r8g8b8, 0x2d + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x11dffff, + PIXMAN_r3g3b2, 0xcb, + PIXMAN_r3g3b2, 0x8 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_r5g6b5, 0xbdab, + PIXMAN_a4r4g4b4, 0xbb0 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff3343, + PIXMAN_a8r8g8b8, 0x7a00ffff, + PIXMAN_a2r2g2b2, 0xd + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ebff4b, + PIXMAN_r3g3b2, 0x26, + PIXMAN_r3g3b2, 0x24 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x2c1b3ff, + PIXMAN_a8r8g8b8, 0x3000152a, + PIXMAN_r3g3b2, 0x24 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1a7ffff, + PIXMAN_r3g3b2, 0x9, + PIXMAN_r5g6b5, 0x24a + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x4ff00ec, + PIXMAN_a8r8g8b8, 0x1da4961e, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff25ff, + PIXMAN_a8r8g8b8, 0x64b0ff00, + PIXMAN_r5g6b5, 0x606c + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1fd62ff, + PIXMAN_a4r4g4b4, 0x76b1, + PIXMAN_r5g6b5, 0x716e + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x194ffde, + PIXMAN_r5g6b5, 0x47ff, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x108ffff, + PIXMAN_a8r8g8b8, 0xffffff66, + PIXMAN_r5g6b5, 0xff0c + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x5ffffff, + PIXMAN_r5g6b5, 0xdf, + PIXMAN_r5g6b5, 0xc0 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ad31, + PIXMAN_a2r2g2b2, 0xc5, + PIXMAN_a4r4g4b4, 0x31 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff34, + PIXMAN_a8r8g8b8, 0x6a57c491, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1fffff1, + PIXMAN_r3g3b2, 0xaf, + PIXMAN_r5g6b5, 0xb01e + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff67ff, + PIXMAN_a4r4g4b4, 0x50ff, + PIXMAN_a8r8g8b8, 0x552255 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x11bffff, + PIXMAN_r5g6b5, 0xef0c, + PIXMAN_r5g6b5, 0xc + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x16cf37d, + PIXMAN_a4r4g4b4, 0xc561, + PIXMAN_r5g6b5, 0x2301 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffff9c, + PIXMAN_a4r4g4b4, 0x2700, + PIXMAN_a8r8g8b8, 0xffff + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x200f322, + PIXMAN_a8r8g8b8, 0xff3c7e, + PIXMAN_r5g6b5, 0x2 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1f14a33, + PIXMAN_a8r8g8b8, 0x26cff79, + PIXMAN_r3g3b2, 0xf9 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x11d922c, + PIXMAN_r3g3b2, 0xab, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_a2r2g2b2, 0xf5, + PIXMAN_r3g3b2, 0x9 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x18697ff, + PIXMAN_a4r4g4b4, 0x5700, + PIXMAN_r5g6b5, 0xfa6d + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x12000fc, + PIXMAN_a2r2g2b2, 0x41, + PIXMAN_a8r8g8b8, 0xb0054 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x100ccff, + PIXMAN_a4r4g4b4, 0x657e, + PIXMAN_r5g6b5, 0x3b1 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffff1f, + PIXMAN_a2r2g2b2, 0xa6, + PIXMAN_r5g6b5, 0x2a0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x11fff82, + PIXMAN_a4r4g4b4, 0xff94, + PIXMAN_a8r8g8b8, 0x1010123 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x154bd19, + PIXMAN_a4r4g4b4, 0xb600, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x10000ff, + PIXMAN_r5g6b5, 0x8e, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x21aff00, + PIXMAN_r5g6b5, 0x71ff, + PIXMAN_r3g3b2, 0xf2 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ad00a7, + PIXMAN_a4r4g4b4, 0x23, + PIXMAN_a8r8g8b8, 0x21 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x100ff00, + PIXMAN_r5g6b5, 0xb343, + PIXMAN_r3g3b2, 0xc + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x3ffa500, + PIXMAN_a8r8g8b8, 0x1af5b4, + PIXMAN_a8r8g8b8, 0xff1abc00 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ffff11, + PIXMAN_a8r8g8b8, 0x9f334f, + PIXMAN_a8r8g8b8, 0x9f0005 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x2c75971, + PIXMAN_a4r4g4b4, 0x3900, + PIXMAN_a4r4g4b4, 0x211 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100ff49, + PIXMAN_a8r8g8b8, 0x813dc25e, + PIXMAN_r5g6b5, 0x667d + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x10000ff, + PIXMAN_a4r4g4b4, 0x4bff, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x20ebcff, + PIXMAN_r5g6b5, 0xc9ff, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_r5g6b5, 0x51ff, + PIXMAN_r3g3b2, 0x44 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffd158, + PIXMAN_a8r8g8b8, 0x7d88ffce, + PIXMAN_r3g3b2, 0x6c + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1425e21, + PIXMAN_a2r2g2b2, 0xa5, + PIXMAN_r5g6b5, 0xe1 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x14b00ff, + PIXMAN_a8r8g8b8, 0xbe95004b, + PIXMAN_r5g6b5, 0x9 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x14fc0cd, + PIXMAN_a8r8g8b8, 0x2d12b78b, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff8230, + PIXMAN_a2r2g2b2, 0x4c, + PIXMAN_r3g3b2, 0x44 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff31ff, + PIXMAN_a2r2g2b2, 0x14, + PIXMAN_a8r8g8b8, 0x551000 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x17800ff, + PIXMAN_a4r4g4b4, 0x22, + PIXMAN_a8r8g8b8, 0x22 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x14500ff, + PIXMAN_a4r4g4b4, 0x6400, + PIXMAN_r5g6b5, 0xff78 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ff9d, + PIXMAN_r3g3b2, 0xcd, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x3ff00ff, + PIXMAN_a4r4g4b4, 0xf269, + PIXMAN_a4r4g4b4, 0x200 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ff28b8, + PIXMAN_a4r4g4b4, 0x33ff, + PIXMAN_r5g6b5, 0x3000 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1006278, + PIXMAN_a8r8g8b8, 0x8a7f18, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffcb00, + PIXMAN_a4r4g4b4, 0x7900, + PIXMAN_a2r2g2b2, 0x14 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x115ff00, + PIXMAN_a8r8g8b8, 0x508d, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x3ff30b5, + PIXMAN_r5g6b5, 0x2e60, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x182fffb, + PIXMAN_r3g3b2, 0x1, + PIXMAN_a8r8g8b8, 0x1000054 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x16fff00, + PIXMAN_r5g6b5, 0x7bc0, + PIXMAN_a8r8g8b8, 0x367900 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1d95dd8, + PIXMAN_a4r4g4b4, 0xfff5, + PIXMAN_r5g6b5, 0xff09 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ff3cdc, + PIXMAN_a8r8g8b8, 0x3bda45ff, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x13900f8, + PIXMAN_a8r8g8b8, 0x7e00ffff, + PIXMAN_a4r4g4b4, 0xff00 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x10ea9ff, + PIXMAN_a8r8g8b8, 0xff34ff22, + PIXMAN_r5g6b5, 0xff52 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2002e99, + PIXMAN_a4r4g4b4, 0x3000, + PIXMAN_r5g6b5, 0x43 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_r5g6b5, 0x19ff, + PIXMAN_r3g3b2, 0x3 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_a8r8g8b8, 0xffff4251, + PIXMAN_a2r2g2b2, 0x4 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x121c9ff, + PIXMAN_a4r4g4b4, 0xd2, + PIXMAN_a4r4g4b4, 0x2 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ff4d, + PIXMAN_a2r2g2b2, 0x5e, + PIXMAN_a2r2g2b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x29ab4ff, + PIXMAN_r3g3b2, 0x47, + PIXMAN_a8r8g8b8, 0x1900 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffc1ac, + PIXMAN_a8r8g8b8, 0xee4ed0ac, + PIXMAN_a8r8g8b8, 0x1009d74 + }, + { PIXMAN_OP_CONJOINT_IN_REVERSE, + PIXMAN_a8r8g8b8, 0x269dffdc, + PIXMAN_a8r8g8b8, 0xff0b00e0, + PIXMAN_a8r8g8b8, 0x2a200ff + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffffff, + PIXMAN_a4r4g4b4, 0x3200, + PIXMAN_r3g3b2, 0x24 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x100ffed, + PIXMAN_a8r8g8b8, 0x67004eff, + PIXMAN_a2r2g2b2, 0x5 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2fffd6a, + PIXMAN_a8r8g8b8, 0xc9003bff, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x253ff00, + PIXMAN_r5g6b5, 0xff, + PIXMAN_r5g6b5, 0xe0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x13600ad, + PIXMAN_r5g6b5, 0x35ae, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffa8ff, + PIXMAN_a8r8g8b8, 0xff5f00, + PIXMAN_r3g3b2, 0xe0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x10067ff, + PIXMAN_a4r4g4b4, 0x450d, + PIXMAN_a2r2g2b2, 0x1 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1ff01ff, + PIXMAN_r3g3b2, 0x77, + PIXMAN_r5g6b5, 0x6800 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x11da4ff, + PIXMAN_r5g6b5, 0x83c9, + PIXMAN_a4r4g4b4, 0x44 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffd4ff, + PIXMAN_r3g3b2, 0xaa, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ff0000, + PIXMAN_a8r8g8b8, 0x71002a, + PIXMAN_a4r4g4b4, 0x700 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1d7ffff, + PIXMAN_r5g6b5, 0x3696, + PIXMAN_a4r4g4b4, 0x200 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffffc8, + PIXMAN_r5g6b5, 0xe900, + PIXMAN_a8r8g8b8, 0x2000 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff004a, + PIXMAN_r3g3b2, 0x48, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x3ffe969, + PIXMAN_r5g6b5, 0xff, + PIXMAN_r5g6b5, 0xc0 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x300ff73, + PIXMAN_r5g6b5, 0xff, + PIXMAN_a8r8g8b8, 0x3000073 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ff93ff, + PIXMAN_a8r8g8b8, 0x61fc7d2b, + PIXMAN_a4r4g4b4, 0x2 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x11bffff, + PIXMAN_a4r4g4b4, 0xffb4, + PIXMAN_r5g6b5, 0x8 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x1e9e100, + PIXMAN_a2r2g2b2, 0x56, + PIXMAN_a2r2g2b2, 0x14 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x3ffb656, + PIXMAN_r3g3b2, 0x4, + PIXMAN_a4r4g4b4, 0xff99 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ff00, + PIXMAN_r3g3b2, 0x68, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1006dff, + PIXMAN_a2r2g2b2, 0x5d, + PIXMAN_a8r8g8b8, 0xff00ff55 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x11c00cb, + PIXMAN_a2r2g2b2, 0x44, + PIXMAN_a4r4g4b4, 0x4 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1d0ff86, + PIXMAN_r3g3b2, 0x5c, + PIXMAN_a8r8g8b8, 0x3c0000 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x2f25fff, + PIXMAN_r3g3b2, 0x36, + PIXMAN_a8r8g8b8, 0x2a444aa + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x134af85, + PIXMAN_r3g3b2, 0x29, + PIXMAN_r5g6b5, 0xf300 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x13398af, + PIXMAN_r3g3b2, 0xa5, + PIXMAN_a4r4g4b4, 0x13 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff57ff, + PIXMAN_a4r4g4b4, 0x252c, + PIXMAN_r3g3b2, 0x40 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x115ffff, + PIXMAN_r5g6b5, 0xffe3, + PIXMAN_r5g6b5, 0x3303 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffff00, + PIXMAN_r5g6b5, 0x6300, + PIXMAN_r3g3b2, 0x6c + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x4ccff9c, + PIXMAN_r5g6b5, 0xcc, + PIXMAN_a8r8g8b8, 0x400003d + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffc6dd, + PIXMAN_r5g6b5, 0x9bff, + PIXMAN_r5g6b5, 0x5bff + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x14fff95, + PIXMAN_r3g3b2, 0x46, + PIXMAN_a8r8g8b8, 0x1000063 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1e6b700, + PIXMAN_r5g6b5, 0xc1ff, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffff54, + PIXMAN_a8r8g8b8, 0x2e00ff, + PIXMAN_r5g6b5, 0x2800 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3ffffff, + PIXMAN_r5g6b5, 0xff, + PIXMAN_r5g6b5, 0xe0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1003550, + PIXMAN_r5g6b5, 0xffcc, + PIXMAN_r5g6b5, 0x1e0 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ffff74, + PIXMAN_r3g3b2, 0x28, + PIXMAN_a8r8g8b8, 0xfe2f49d7 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1e35100, + PIXMAN_r3g3b2, 0x57, + PIXMAN_r5g6b5, 0x4000 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x268ffa3, + PIXMAN_a4r4g4b4, 0x30, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x35700f8, + PIXMAN_r5g6b5, 0xa4, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x3ce1dff, + PIXMAN_r5g6b5, 0x2a5e, + PIXMAN_a8r8g8b8, 0x210000 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x494a7ff, + PIXMAN_a8r8g8b8, 0x1bffe400, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x10026d9, + PIXMAN_a8r8g8b8, 0xec00621f, + PIXMAN_r5g6b5, 0x63 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ff99, + PIXMAN_a8r8g8b8, 0xf334ff, + PIXMAN_a4r4g4b4, 0x30 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffc200, + PIXMAN_a8r8g8b8, 0x1e0000ff, + PIXMAN_a8r8g8b8, 0x1e1700 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff00ff, + PIXMAN_r3g3b2, 0x4b, + PIXMAN_r5g6b5, 0x4818 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x2e800ff, + PIXMAN_a4r4g4b4, 0xd3, + PIXMAN_a4r4g4b4, 0xec + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x19a001f, + PIXMAN_r3g3b2, 0x76, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1cb00c3, + PIXMAN_a4r4g4b4, 0x5cff, + PIXMAN_r5g6b5, 0x4008 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff0000, + PIXMAN_r3g3b2, 0x2a, + PIXMAN_r5g6b5, 0xc5fb + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_a8r8g8b8, 0xea005a88, + PIXMAN_r3g3b2, 0xb3 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100ffea, + PIXMAN_a4r4g4b4, 0x54eb, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x179ffff, + PIXMAN_r3g3b2, 0xa4, + PIXMAN_a8r8g8b8, 0x2400 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x17ad226, + PIXMAN_r3g3b2, 0xa4, + PIXMAN_r5g6b5, 0xe0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ff01, + PIXMAN_a2r2g2b2, 0x25, + PIXMAN_a4r4g4b4, 0x50 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x20000ff, + PIXMAN_a8r8g8b8, 0x2b00c127, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x200ff96, + PIXMAN_a4r4g4b4, 0x2300, + PIXMAN_r3g3b2, 0x6 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x200ffff, + PIXMAN_r3g3b2, 0x87, + PIXMAN_r5g6b5, 0x5bc8 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1fffff2, + PIXMAN_r3g3b2, 0x7e, + PIXMAN_a2r2g2b2, 0xe + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1ff8b00, + PIXMAN_a4r4g4b4, 0xd500, + PIXMAN_r3g3b2, 0x40 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ffffff, + PIXMAN_a8r8g8b8, 0x1bff38, + PIXMAN_a4r4g4b4, 0xf0 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x158ff39, + PIXMAN_a4r4g4b4, 0x75dd, + PIXMAN_a8r8g8b8, 0xdd31 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1009b70, + PIXMAN_a4r4g4b4, 0xff40, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x12fb43f, + PIXMAN_a4r4g4b4, 0x69ff, + PIXMAN_a2r2g2b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff95, + PIXMAN_a2r2g2b2, 0x84, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x200d188, + PIXMAN_r5g6b5, 0xde6, + PIXMAN_r5g6b5, 0x3 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2c70000, + PIXMAN_r5g6b5, 0x24fa, + PIXMAN_a8r8g8b8, 0x21a0000 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x100ff24, + PIXMAN_a4r4g4b4, 0x835, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x10000cd, + PIXMAN_a2r2g2b2, 0x7f, + PIXMAN_a2r2g2b2, 0x1 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x379ffff, + PIXMAN_a8r8g8b8, 0x23ffff00, + PIXMAN_r5g6b5, 0x4eda + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x172e3ff, + PIXMAN_r3g3b2, 0xa6, + PIXMAN_r5g6b5, 0x100 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100f5ad, + PIXMAN_a4r4g4b4, 0x7908, + PIXMAN_a2r2g2b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100fff9, + PIXMAN_a2r2g2b2, 0xf1, + PIXMAN_r3g3b2, 0x1 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1abff00, + PIXMAN_r5g6b5, 0x31ff, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x112ffd1, + PIXMAN_r3g3b2, 0x9, + PIXMAN_a2r2g2b2, 0xdd + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ffbf, + PIXMAN_r3g3b2, 0x2c, + PIXMAN_a4r4g4b4, 0x60 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffb7ff, + PIXMAN_r3g3b2, 0x6b, + PIXMAN_a4r4g4b4, 0x630 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x20005ff, + PIXMAN_a4r4g4b4, 0x8462, + PIXMAN_r5g6b5, 0xb1e8 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff5b00, + PIXMAN_r5g6b5, 0x70ff, + PIXMAN_r3g3b2, 0x60 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffffc3, + PIXMAN_r3g3b2, 0x39, + PIXMAN_a8r8g8b8, 0x200db41 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x306ffff, + PIXMAN_a8r8g8b8, 0xdcffff1f, + PIXMAN_a8r8g8b8, 0x306ff00 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x193daff, + PIXMAN_a8r8g8b8, 0x69000000, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x2a200ff, + PIXMAN_a8r8g8b8, 0x183aff00, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100f1a5, + PIXMAN_a8r8g8b8, 0xb5fc21ff, + PIXMAN_r5g6b5, 0xfe00 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1630019, + PIXMAN_a8r8g8b8, 0x6affc400, + PIXMAN_r5g6b5, 0x56ff + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff8bc2, + PIXMAN_r3g3b2, 0xee, + PIXMAN_r5g6b5, 0x1c0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x260ffff, + PIXMAN_a4r4g4b4, 0x3f00, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x169ffed, + PIXMAN_a8r8g8b8, 0xffffff3f, + PIXMAN_a8r8g8b8, 0x169ff00 + }, + { PIXMAN_OP_CONJOINT_XOR, + PIXMAN_a8r8g8b8, 0x154c181, + PIXMAN_a4r4g4b4, 0x5100, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1e09c00, + PIXMAN_r5g6b5, 0xca00, + PIXMAN_a4r4g4b4, 0xb00 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ff8dff, + PIXMAN_a8r8g8b8, 0x610038ff, + PIXMAN_a8r8g8b8, 0x1001f02 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1e400ff, + PIXMAN_a4r4g4b4, 0x66bd, + PIXMAN_r3g3b2, 0x68 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x25362ff, + PIXMAN_a4r4g4b4, 0x31ff, + PIXMAN_a8r8g8b8, 0x111433 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3ad0039, + PIXMAN_r3g3b2, 0x26, + PIXMAN_a8r8g8b8, 0x3000026 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2e442ef, + PIXMAN_r3g3b2, 0x32, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1720000, + PIXMAN_a8r8g8b8, 0x55fdea00, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x14bb0d7, + PIXMAN_a8r8g8b8, 0x7fffff47, + PIXMAN_a2r2g2b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x13dffff, + PIXMAN_a8r8g8b8, 0xa3860672, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x120495a, + PIXMAN_a4r4g4b4, 0x407e, + PIXMAN_a8r8g8b8, 0x54 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff8fff, + PIXMAN_a2r2g2b2, 0x29, + PIXMAN_r5g6b5, 0xa + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100a31a, + PIXMAN_a4r4g4b4, 0xde4c, + PIXMAN_a4r4g4b4, 0x1 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1d4008c, + PIXMAN_r3g3b2, 0x79, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ff0000, + PIXMAN_a4r4g4b4, 0x7de4, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1b27e62, + PIXMAN_a4r4g4b4, 0x7941, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x300ff00, + PIXMAN_a8r8g8b8, 0xfcff255e, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x2ff00b8, + PIXMAN_a8r8g8b8, 0x19ff718d, + PIXMAN_r5g6b5, 0x1802 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x235ff13, + PIXMAN_a8r8g8b8, 0x34bcd9ff, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1006400, + PIXMAN_a4r4g4b4, 0x7000, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff8bff, + PIXMAN_a4r4g4b4, 0xfff4, + PIXMAN_a4r4g4b4, 0xf80 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x24630ff, + PIXMAN_a8r8g8b8, 0x1f00000b, + PIXMAN_a8r8g8b8, 0x9061f + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff8a00, + PIXMAN_a8r8g8b8, 0x79ffab00, + PIXMAN_r5g6b5, 0x7a00 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x19807ff, + PIXMAN_a4r4g4b4, 0x6794, + PIXMAN_a8r8g8b8, 0xff002e00 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x10000da, + PIXMAN_a4r4g4b4, 0xf864, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffffde, + PIXMAN_a2r2g2b2, 0x94, + PIXMAN_a8r8g8b8, 0x1000000 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x200c800, + PIXMAN_r5g6b5, 0xe9d4, + PIXMAN_a8r8g8b8, 0x2c00 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff00c9, + PIXMAN_r3g3b2, 0x4c, + PIXMAN_r5g6b5, 0x4800 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x122d5ff, + PIXMAN_r5g6b5, 0x418b, + PIXMAN_a4r4g4b4, 0x25 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ffff55, + PIXMAN_a2r2g2b2, 0x1c, + PIXMAN_a8r8g8b8, 0xff00 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x135ffff, + PIXMAN_r5g6b5, 0x39c4, + PIXMAN_r5g6b5, 0xb7 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x100d2c3, + PIXMAN_r3g3b2, 0x2a, + PIXMAN_a8r8g8b8, 0x3c00 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x17268ff, + PIXMAN_a8r8g8b8, 0x7c00ffff, + PIXMAN_r5g6b5, 0x318f + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff00ff, + PIXMAN_r3g3b2, 0x68, + PIXMAN_r3g3b2, 0xb4 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x200ffff, + PIXMAN_r5g6b5, 0xff86, + PIXMAN_a8r8g8b8, 0x200f300 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x18a23ff, + PIXMAN_a2r2g2b2, 0x44, + PIXMAN_a4r4g4b4, 0x205 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x16bff23, + PIXMAN_a8r8g8b8, 0x31fd00ff, + PIXMAN_r3g3b2, 0x7 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x137d1ff, + PIXMAN_a4r4g4b4, 0x56c1, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff5bff, + PIXMAN_a4r4g4b4, 0xfff4, + PIXMAN_a4r4g4b4, 0xf50 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x15c6b00, + PIXMAN_a8r8g8b8, 0x7d008a, + PIXMAN_a4r4g4b4, 0x200 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x12091ff, + PIXMAN_a8r8g8b8, 0xb74cff6b, + PIXMAN_a2r2g2b2, 0x8 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff5bff, + PIXMAN_a8r8g8b8, 0xff6ddce8, + PIXMAN_a2r2g2b2, 0x10 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_a4r4g4b4, 0xffb7, + PIXMAN_a4r4g4b4, 0xb0 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x173ffff, + PIXMAN_r5g6b5, 0xff2c, + PIXMAN_a4r4g4b4, 0x6 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x17102ff, + PIXMAN_a8r8g8b8, 0x955bff66, + PIXMAN_a8r8g8b8, 0x280066 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x3c7ff24, + PIXMAN_r5g6b5, 0xc4, + PIXMAN_r5g6b5, 0x163 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100c2a6, + PIXMAN_r5g6b5, 0xa9b9, + PIXMAN_a4r4g4b4, 0x8 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x26049ff, + PIXMAN_a4r4g4b4, 0xb2, + PIXMAN_r5g6b5, 0x8904 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2f100ff, + PIXMAN_r3g3b2, 0x30, + PIXMAN_a8r8g8b8, 0x2220100 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ffff88, + PIXMAN_r3g3b2, 0x7e, + PIXMAN_r3g3b2, 0x60 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x153ffab, + PIXMAN_a8r8g8b8, 0xfd10725a, + PIXMAN_r3g3b2, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff00d2, + PIXMAN_r5g6b5, 0xff6b, + PIXMAN_a8r8g8b8, 0x101014a + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x100d965, + PIXMAN_a8r8g8b8, 0xff007b00, + PIXMAN_r3g3b2, 0xc + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ec0000, + PIXMAN_r5g6b5, 0x6fff, + PIXMAN_r5g6b5, 0x6000 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x19d59a2, + PIXMAN_a8r8g8b8, 0x4a00ff7a, + PIXMAN_a8r8g8b8, 0x2e1a2f + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1eb0000, + PIXMAN_a4r4g4b4, 0x72bc, + PIXMAN_r5g6b5, 0x1800 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_a4r4g4b4, 0xc034, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x195ff15, + PIXMAN_a4r4g4b4, 0xb7b1, + PIXMAN_r5g6b5, 0x4000 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffdf94, + PIXMAN_a4r4g4b4, 0x78, + PIXMAN_r3g3b2, 0xc + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x26f00ff, + PIXMAN_a4r4g4b4, 0xff93, + PIXMAN_r5g6b5, 0x1dd2 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x2ff3fc5, + PIXMAN_r3g3b2, 0x2f, + PIXMAN_a8r8g8b8, 0x240000 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1ff696e, + PIXMAN_a4r4g4b4, 0x22ff, + PIXMAN_r5g6b5, 0x34d + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x10033d9, + PIXMAN_a8r8g8b8, 0x38650000, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffff00, + PIXMAN_a4r4g4b4, 0x2070, + PIXMAN_r5g6b5, 0x2100 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1008746, + PIXMAN_a8r8g8b8, 0xb56971, + PIXMAN_r5g6b5, 0xc25c + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x144d200, + PIXMAN_a4r4g4b4, 0xff42, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1ffffd0, + PIXMAN_r5g6b5, 0x5b00, + PIXMAN_r3g3b2, 0x4c + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x10000ff, + PIXMAN_a8r8g8b8, 0xff006f, + PIXMAN_r5g6b5, 0xd + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x10666ff, + PIXMAN_a4r4g4b4, 0x39b2, + PIXMAN_r5g6b5, 0xa6 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x11a007d, + PIXMAN_r3g3b2, 0xf9, + PIXMAN_a8r8g8b8, 0x11a0000 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1eb90ee, + PIXMAN_r5g6b5, 0xd, + PIXMAN_a2r2g2b2, 0x1 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ff42d5, + PIXMAN_a4r4g4b4, 0x3400, + PIXMAN_r3g3b2, 0x40 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1dfff00, + PIXMAN_a8r8g8b8, 0x3ffff9d2, + PIXMAN_r5g6b5, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff6500, + PIXMAN_a2r2g2b2, 0x56, + PIXMAN_r3g3b2, 0x44 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x119ffe6, + PIXMAN_r3g3b2, 0x8d, + PIXMAN_a4r4g4b4, 0xff00 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x100cd00, + PIXMAN_r5g6b5, 0x33ff, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x569ffd7, + PIXMAN_r5g6b5, 0x8cc, + PIXMAN_r5g6b5, 0xc0 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100876a, + PIXMAN_a8r8g8b8, 0x575447a5, + PIXMAN_r5g6b5, 0x164 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x12d00ff, + PIXMAN_a4r4g4b4, 0x3fff, + PIXMAN_a4r4g4b4, 0x0 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x2ff953b, + PIXMAN_a4r4g4b4, 0x2914, + PIXMAN_r5g6b5, 0x20a1 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffead4, + PIXMAN_a8r8g8b8, 0xff00ea4e, + PIXMAN_r3g3b2, 0x5a + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x1ff6400, + PIXMAN_a2r2g2b2, 0x99, + PIXMAN_r5g6b5, 0xa620 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x17b0084, + PIXMAN_r3g3b2, 0xbd, + PIXMAN_a4r4g4b4, 0x500 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x4f90bbb, + PIXMAN_a8r8g8b8, 0xff00d21f, + PIXMAN_a8r8g8b8, 0xfb00fc4a + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ffbb1d, + PIXMAN_a8r8g8b8, 0x2dff79ff, + PIXMAN_r5g6b5, 0x2c0 + }, + { PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ffff, + PIXMAN_a2r2g2b2, 0x43, + PIXMAN_a4r4g4b4, 0x6f + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1f000ff, + PIXMAN_a4r4g4b4, 0xb393, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1c60020, + PIXMAN_a8r8g8b8, 0x6bffffff, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1727d00, + PIXMAN_a2r2g2b2, 0x67, + PIXMAN_a4r4g4b4, 0x400 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x14a5194, + PIXMAN_a4r4g4b4, 0xd7ff, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x20003fa, + PIXMAN_a4r4g4b4, 0x24ff, + PIXMAN_a8r8g8b8, 0xffff1550 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1a6ff83, + PIXMAN_a4r4g4b4, 0xf400, + PIXMAN_r5g6b5, 0x2800 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ffcf00, + PIXMAN_r5g6b5, 0x71ff, + PIXMAN_a4r4g4b4, 0x30 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x333ffff, + PIXMAN_a4r4g4b4, 0x2c00, + PIXMAN_r3g3b2, 0x4 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1c2ffe8, + PIXMAN_r5g6b5, 0xc200, + PIXMAN_a8r8g8b8, 0xfeca41ff + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a2r2g2b2, 0x47, + PIXMAN_a8r8g8b8, 0x2ffff00, + PIXMAN_a8r8g8b8, 0x3aa0102 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffeb00, + PIXMAN_a4r4g4b4, 0xb493, + PIXMAN_a4r4g4b4, 0x400 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2afffff, + PIXMAN_r5g6b5, 0xcb, + PIXMAN_r5g6b5, 0xc0 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x183ff00, + PIXMAN_r3g3b2, 0x87, + PIXMAN_r5g6b5, 0xae91 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x3ffff00, + PIXMAN_a4r4g4b4, 0x2ba4, + PIXMAN_r5g6b5, 0x2100 + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x215cbc2, + PIXMAN_a4r4g4b4, 0xafd3, + PIXMAN_a8r8g8b8, 0x115b000 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1853f65, + PIXMAN_a8r8g8b8, 0xc68cdc41, + PIXMAN_r5g6b5, 0x3 + }, + { PIXMAN_OP_CONJOINT_IN, + PIXMAN_a8r8g8b8, 0x3ffff8f, + PIXMAN_a4r4g4b4, 0x8824, + PIXMAN_a4r4g4b4, 0x20 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x28e08e6, + PIXMAN_a8r8g8b8, 0x2cffff31, + PIXMAN_r5g6b5, 0x1805 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x1b500be, + PIXMAN_r5g6b5, 0xd946, + PIXMAN_r5g6b5, 0x9800 + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x133ffb3, + PIXMAN_a2r2g2b2, 0x42, + PIXMAN_a8r8g8b8, 0x11553c + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x21aff81, + PIXMAN_r3g3b2, 0xc7, + PIXMAN_r5g6b5, 0x120 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x12e004f, + PIXMAN_a4r4g4b4, 0xf617, + PIXMAN_a4r4g4b4, 0x102 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x164861f, + PIXMAN_r3g3b2, 0x4e, + PIXMAN_r5g6b5, 0x19c0 + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff0eff, + PIXMAN_a8r8g8b8, 0xff5c00aa, + PIXMAN_r5g6b5, 0x5800 + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x1e4c60f, + PIXMAN_a8r8g8b8, 0x38ff0e0c, + PIXMAN_a4r4g4b4, 0xff2a + }, + { PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff0000, + PIXMAN_a8r8g8b8, 0x9f3d6700, + PIXMAN_r5g6b5, 0xf3ff + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x205ffd0, + PIXMAN_a8r8g8b8, 0xffc22b3b, + PIXMAN_a8r8g8b8, 0x2040000 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x1ff0059, + PIXMAN_r5g6b5, 0x74ff, + PIXMAN_a8r8g8b8, 0x1730101 + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x29affb8, + PIXMAN_r5g6b5, 0xff, + PIXMAN_a8r8g8b8, 0x2d25cff + }, + { PIXMAN_OP_DISJOINT_OUT, + PIXMAN_a8r8g8b8, 0x1ffff8b, + PIXMAN_a4r4g4b4, 0xff7b, + PIXMAN_r5g6b5, 0x3a0 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x2a86ad7, + PIXMAN_a4r4g4b4, 0xdc22, + PIXMAN_a8r8g8b8, 0x2860000 + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x3ff00ff, + PIXMAN_r3g3b2, 0x33, + PIXMAN_r5g6b5, 0x2000 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1e50063, + PIXMAN_a8r8g8b8, 0x35ff95d7, + PIXMAN_r3g3b2, 0x20 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x2ffe6ff, + PIXMAN_a8r8g8b8, 0x153ef297, + PIXMAN_r5g6b5, 0x6d2 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x34ffeff, + PIXMAN_a4r4g4b4, 0x2e, + PIXMAN_r5g6b5, 0x1d + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2ffeada, + PIXMAN_r5g6b5, 0xabc6, + PIXMAN_a8r8g8b8, 0xfd15b256 + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x100ff00, + PIXMAN_a8r8g8b8, 0xcff3f32, + PIXMAN_a8r8g8b8, 0x3f00 + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x1e1b0f1, + PIXMAN_a8r8g8b8, 0xff63ff54, + PIXMAN_r3g3b2, 0x5d + }, + { PIXMAN_OP_DISJOINT_XOR, + PIXMAN_a8r8g8b8, 0x2ffff23, + PIXMAN_a8r8g8b8, 0x380094ff, + PIXMAN_r5g6b5, 0x3a4b + }, + { PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_a4r4g4b4, 0x1000, + PIXMAN_r5g6b5, 0xca, + PIXMAN_a8r8g8b8, 0x3434500 + }, + { PIXMAN_OP_DISJOINT_IN, + PIXMAN_a8r8g8b8, 0x195ffe5, + PIXMAN_a4r4g4b4, 0x3a29, + PIXMAN_a8r8g8b8, 0x0 + }, + { PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_a8r8g8b8, 0x139007a, + PIXMAN_a4r4g4b4, 0x4979, + PIXMAN_r5g6b5, 0x84 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa9, + PIXMAN_a4r4g4b4, 0xfa18, + PIXMAN_a8r8g8b8, 0xabff67ff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x94, + PIXMAN_a4r4g4b4, 0x5109, + PIXMAN_a8r8g8b8, 0x3affffff + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_r5g6b5, 0xd038, + PIXMAN_r5g6b5, 0xff00, + PIXMAN_r5g6b5, 0xf9a5 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0x543128ff, + PIXMAN_a8r8g8b8, 0x7029ff, + PIXMAN_a8r8g8b8, 0x316b1d7 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_r5g6b5, 0x53ff, + PIXMAN_r5g6b5, 0x72ff, + PIXMAN_a8r8g8b8, 0xffffdeff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0x5b00002b, + PIXMAN_a4r4g4b4, 0xc3, + PIXMAN_a8r8g8b8, 0x23530be + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0xcefc0041, + PIXMAN_a8r8g8b8, 0xf60d02, + PIXMAN_a8r8g8b8, 0x1f2ffe5 + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_r5g6b5, 0xffdb, + PIXMAN_r5g6b5, 0xc700, + PIXMAN_r5g6b5, 0x654 + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_r5g6b5, 0xffc6, + PIXMAN_r5g6b5, 0xff09, + PIXMAN_r5g6b5, 0xfe58 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x95, + PIXMAN_r5g6b5, 0x1b4a, + PIXMAN_a8r8g8b8, 0xab234cff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x95, + PIXMAN_a4r4g4b4, 0x5e99, + PIXMAN_a8r8g8b8, 0x3b1c1cdd + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_r5g6b5, 0x22, + PIXMAN_r5g6b5, 0xd00, + PIXMAN_r5g6b5, 0xfbb1 + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_r5g6b5, 0xffc8, + PIXMAN_a8r8g8b8, 0xa1a3ffff, + PIXMAN_r5g6b5, 0x44a + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0xffff7cff, + PIXMAN_r5g6b5, 0x900, + PIXMAN_a8r8g8b8, 0xffff94ec + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xa7, + PIXMAN_r5g6b5, 0xff, + PIXMAN_a8r8g8b8, 0xaa00cffe + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0x85, + PIXMAN_r5g6b5, 0xffb3, + PIXMAN_a8r8g8b8, 0xaaffff4a + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a8r8g8b8, 0x3500a118, + PIXMAN_a4r4g4b4, 0x9942, + PIXMAN_a8r8g8b8, 0x01ff405e + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xb5, + PIXMAN_x4a4, 0xe, + PIXMAN_a8r8g8b8, 0xffbaff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xe872, + PIXMAN_x2r10g10b10, 0xa648ff00, + PIXMAN_a2r10g10b10, 0x14ff00e8, + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x4d2db34, + PIXMAN_a8, 0x19, + PIXMAN_r5g6b5, 0x9700, + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x2ff0076, + PIXMAN_a8r8g8b8, 0x2a0000, + PIXMAN_r3g3b2, 0x0, + }, + { PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_a8r8g8b8, 0x14f00ff, + PIXMAN_r5g6b5, 0xd48, + PIXMAN_a4r4g4b4, 0x0, + }, + { PIXMAN_OP_CONJOINT_OUT, + PIXMAN_a8r8g8b8, 0x3d8bbff, + PIXMAN_r5g6b5, 0x6900, + PIXMAN_a8r8g8b8, 0x0, + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x2ff00ff, + PIXMAN_a4r4g4b4, 0x2300, + PIXMAN_r3g3b2, 0x0, + }, + { PIXMAN_OP_SATURATE, + PIXMAN_a8r8g8b8, 0x4d2db34, + PIXMAN_a8r8g8b8, 0xff0019ff, + PIXMAN_r5g6b5, 0x9700, + }, + { PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_a8r8g8b8, 0x100ac05, + PIXMAN_r3g3b2, 0xef, + PIXMAN_a2r2g2b2, 0xff, + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a2r2g2b2, 0xbf, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x7e + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r5g6b5, 0xffff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x33 + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_a8r8g8b8, 0x84c4ffd7, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xffddff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a8r8g8b8, 0xff6e56, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x20ff1ade + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a4r4g4b4, 0xfe0, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xbdff + }, + { PIXMAN_OP_SCREEN, + PIXMAN_a8r8g8b8, 0x9671ff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x43 + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a2r2g2b2, 0xff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x39ff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_r5g6b5, 0xffff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x1968 + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a4r4g4b4, 0x4247, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xd8ffff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_r5g6b5, 0xff00, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x79 + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r3g3b2, 0xe0, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x39 + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a8r8g8b8, 0xfff8, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0xff + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_r5g6b5, 0x75fc, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x11ff, + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_r3g3b2, 0x52, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0xc627 + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0x9f2b, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x4b00e7f5 + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a8r8g8b8, 0x00dfff5c, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x5e0f, + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_a8r8g8b8, 0xff00121b, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x3776 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_r5g6b5, 0x03e0, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x01003c00, + }, { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, 0x0f00c300, + PIXMAN_null, 0x00, PIXMAN_x14r6g6b6, 0x003c0, }, { PIXMAN_OP_DISJOINT_XOR, PIXMAN_a4r4g4b4, 0xd0c0, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x5300ea00, }, { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, 0x20c6bf00, + PIXMAN_null, 0x00, PIXMAN_r5g6b5, 0xb9ff }, { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, 0x204ac7ff, + PIXMAN_null, 0x00, PIXMAN_r5g6b5, 0xc1ff }, { PIXMAN_OP_OVER_REVERSE, PIXMAN_r5g6b5, 0xffc3, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x102d00dd }, { PIXMAN_OP_OVER_REVERSE, PIXMAN_r5g6b5, 0x1f00, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x1bdf0c89 }, { PIXMAN_OP_OVER_REVERSE, PIXMAN_r5g6b5, 0xf9d2, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x1076bcf7 }, { PIXMAN_OP_OVER_REVERSE, PIXMAN_r5g6b5, 0x00c3, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x1bfe9ae5 }, { PIXMAN_OP_OVER_REVERSE, PIXMAN_r5g6b5, 0x09ff, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x0b00c16c }, { PIXMAN_OP_DISJOINT_ATOP, PIXMAN_a2r2g2b2, 0xbc, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x9efff1ff }, { PIXMAN_OP_DISJOINT_ATOP, PIXMAN_a4r4g4b4, 0xae5f, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0xf215b675 }, { PIXMAN_OP_DISJOINT_ATOP_REVERSE, PIXMAN_a8r8g8b8, 0xce007980, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x80ffe4ad }, { PIXMAN_OP_DISJOINT_XOR, PIXMAN_a8r8g8b8, 0xb8b07bea, + PIXMAN_null, 0x00, PIXMAN_a4r4g4b4, 0x939c }, { PIXMAN_OP_CONJOINT_ATOP_REVERSE, PIXMAN_r5g6b5, 0x0063, + PIXMAN_null, 0x00, PIXMAN_a8r8g8b8, 0x10bb1ed7, }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a2r2g2b2, 0xbf, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x7e + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a8r8g8b8, 0xffffff, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xff3fffff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_r3g3b2, 0x38, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x5b + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_a8r8g8b8, 0x2e9effff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x77 + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r5g6b5, 0xffff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x33 + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a8r8g8b8, 0xd0089ff, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0xb1 + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_r3g3b2, 0x8a, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xcd0004 + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_a8r8g8b8, 0xffff1e3a, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xcf00 + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_a8r8g8b8, 0x84c4ffd7, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xffddff + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_a4r4g4b4, 0xfd75, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x7f + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_r3g3b2, 0xff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x63ff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a8r8g8b8, 0xff6e56, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x20ff1ade + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a4r4g4b4, 0xfe0, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xbdff + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_r5g6b5, 0x9799, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x8d + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_a8r8g8b8, 0xe8ff1c33, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0x6200 + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_a8r8g8b8, 0x22ffffff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x63 + }, + { PIXMAN_OP_SCREEN, + PIXMAN_a8r8g8b8, 0x9671ff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x43 + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a2r2g2b2, 0x83, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0xff + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_r3g3b2, 0x0, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x97 + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_r5g6b5, 0xb900, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x6800ff00 + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a4r4g4b4, 0xff, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0x8e + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xff00, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0xbc + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r5g6b5, 0xfffe, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x90 + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_r3g3b2, 0xff, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xc35f + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a2r2g2b2, 0xff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x39ff + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a2r2g2b2, 0x1e, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xbaff + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a8r8g8b8, 0xb4ffff26, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0xff + }, + { PIXMAN_OP_COLOR_DODGE, + PIXMAN_a4r4g4b4, 0xe3ff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x878b + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a8r8g8b8, 0xff700044, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x6 + }, + { PIXMAN_OP_DARKEN, + PIXMAN_a2r2g2b2, 0xb6, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xcd00 + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_a2r2g2b2, 0xfe, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x12 + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a8r8g8b8, 0xb1ff006c, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xff7c + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r3g3b2, 0x4e, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x3c + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_r5g6b5, 0xffff, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0x1968 + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_r3g3b2, 0xe7, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x8cced6ac + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a4r4g4b4, 0xa500, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x1bff009d + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r5g6b5, 0x45ff, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x32 + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a2r2g2b2, 0x18, + PIXMAN_null, 0x00, + PIXMAN_r5g6b5, 0xdc00 + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a4r4g4b4, 0x4247, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xd8ffff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_r5g6b5, 0xff00, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x79 + }, + { PIXMAN_OP_COLOR_BURN, + PIXMAN_r3g3b2, 0xf, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x9fff00ff + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a2r2g2b2, 0x93, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xff + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a2r2g2b2, 0xa3, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0xca + }, + { PIXMAN_OP_DIFFERENCE, + PIXMAN_r3g3b2, 0xe0, + PIXMAN_null, 0x00, + PIXMAN_a2r2g2b2, 0x39 + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r3g3b2, 0x16, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x98ffff + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_r3g3b2, 0x96, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0x225f6c + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_a4r4g4b4, 0x12c7, + PIXMAN_null, 0x00, + PIXMAN_a4r4g4b4, 0xb100 + }, + { PIXMAN_OP_LIGHTEN, + PIXMAN_a8r8g8b8, 0xffda91, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0x6a + }, + { PIXMAN_OP_EXCLUSION, + PIXMAN_a8r8g8b8, 0xfff8, + PIXMAN_null, 0x00, + PIXMAN_r3g3b2, 0xff + }, + { PIXMAN_OP_SOFT_LIGHT, + PIXMAN_a2r2g2b2, 0xff, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xf0ff48ca + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xf1ff, + PIXMAN_r5g6b5, 0x6eff, + PIXMAN_a8r8g8b8, 0xffffff, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xf1ff, + PIXMAN_a8, 0xdf, + PIXMAN_a8r8g8b8, 0xffffff, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xf1ff, + PIXMAN_null, 0x00, + PIXMAN_a8r8g8b8, 0xffffff, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xb867, + PIXMAN_a4r4g4b4, 0x82d9, + PIXMAN_a8r8g8b8, 0xffc5, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xa9f5, + PIXMAN_r5g6b5, 0xadff, + PIXMAN_a8r8g8b8, 0xffff00, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0x4900, + PIXMAN_r5g6b5, 0x865c, + PIXMAN_a8r8g8b8, 0xebff, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xd9ff, + PIXMAN_a8r8g8b8, 0xffffffff, + PIXMAN_a8r8g8b8, 0x8ff0d, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0x41ff, + PIXMAN_a4r4g4b4, 0xcff, + PIXMAN_a8r8g8b8, 0xe1ff00, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0x91ff, + PIXMAN_a2r2g2b2, 0xf3, + PIXMAN_a8r8g8b8, 0xe4ffb4, + }, + { PIXMAN_OP_HARD_LIGHT, + PIXMAN_r5g6b5, 0xb9ff, + PIXMAN_a2r2g2b2, 0xff, + PIXMAN_a8r8g8b8, 0xffff, + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a8r8g8b8, 0x473affff, + PIXMAN_r5g6b5, 0x2b00, + PIXMAN_r5g6b5, 0x1ff, + }, + { PIXMAN_OP_OVERLAY, + PIXMAN_a8r8g8b8, 0xe4ff, + PIXMAN_r3g3b2, 0xff, + PIXMAN_r5g6b5, 0x89ff, + }, }; static void @@ -159,35 +2855,59 @@ access (pixman_image_t *image, int x, int y) } static pixman_bool_t -verify (int test_no, const pixel_combination_t *combination, int size) +verify (int test_no, const pixel_combination_t *combination, int size, + pixman_bool_t component_alpha) { - pixman_image_t *src, *dest; - pixel_checker_t src_checker, dest_checker; - color_t source_color, dest_color, reference_color; + pixman_image_t *src, *mask, *dest; + pixel_checker_t src_checker, mask_checker, dest_checker; + color_t source_color, mask_color, dest_color, reference_color; + pixman_bool_t have_mask = (combination->mask_format != PIXMAN_null); pixman_bool_t result = TRUE; int i, j; /* Compute reference color */ pixel_checker_init (&src_checker, combination->src_format); + if (have_mask) + pixel_checker_init (&mask_checker, combination->mask_format); pixel_checker_init (&dest_checker, combination->dest_format); + pixel_checker_convert_pixel_to_color ( &src_checker, combination->src_pixel, &source_color); + if (combination->mask_format != PIXMAN_null) + { + pixel_checker_convert_pixel_to_color ( + &mask_checker, combination->mask_pixel, &mask_color); + } pixel_checker_convert_pixel_to_color ( &dest_checker, combination->dest_pixel, &dest_color); + do_composite (combination->op, - &source_color, NULL, &dest_color, - &reference_color, FALSE); + &source_color, + have_mask? &mask_color : NULL, + &dest_color, + &reference_color, component_alpha); src = pixman_image_create_bits ( combination->src_format, size, size, NULL, -1); + if (have_mask) + { + mask = pixman_image_create_bits ( + combination->mask_format, size, size, NULL, -1); + + pixman_image_set_component_alpha (mask, component_alpha); + } dest = pixman_image_create_bits ( combination->dest_format, size, size, NULL, -1); fill (src, combination->src_pixel); + if (have_mask) + fill (mask, combination->mask_pixel); fill (dest, combination->dest_pixel); pixman_image_composite32 ( - combination->op, src, NULL, dest, 0, 0, 0, 0, 0, 0, size, size); + combination->op, src, + have_mask ? mask : NULL, + dest, 0, 0, 0, 0, 0, 0, size, size); for (j = 0; j < size; ++j) { @@ -200,9 +2920,13 @@ verify (int test_no, const pixel_combination_t *combination, int size) { printf ("----------- Test %d failed ----------\n", test_no); - printf (" operator: %s\n", operator_name (combination->op)); + printf (" operator: %s (%s)\n", operator_name (combination->op), + have_mask? component_alpha ? "component alpha" : "unified alpha" : "no mask"); printf (" src format: %s\n", format_name (combination->src_format)); + if (have_mask != PIXMAN_null) + printf (" mask format: %s\n", format_name (combination->mask_format)); printf (" dest format: %s\n", format_name (combination->dest_format)); + printf (" - source ARGB: %f %f %f %f (pixel: %8x)\n", source_color.a, source_color.r, source_color.g, source_color.b, combination->src_pixel); @@ -210,6 +2934,16 @@ verify (int test_no, const pixel_combination_t *combination, int size) &a, &r, &g, &b); printf (" %8d %8d %8d %8d\n", a, r, g, b); + if (have_mask) + { + printf (" - mask ARGB: %f %f %f %f (pixel: %8x)\n", + mask_color.a, mask_color.r, mask_color.g, mask_color.b, + combination->mask_pixel); + pixel_checker_split_pixel (&mask_checker, combination->mask_pixel, + &a, &r, &g, &b); + printf (" %8d %8d %8d %8d\n", a, r, g, b); + } + printf (" - dest ARGB: %f %f %f %f (pixel: %8x)\n", dest_color.a, dest_color.r, dest_color.g, dest_color.b, combination->dest_pixel); @@ -248,19 +2982,41 @@ main (int argc, char **argv) { int result = 0; int i, j; + int lo, hi; + + if (argc > 1) + { + lo = atoi (argv[1]); + hi = lo + 1; + } + else + { + lo = 0; + hi = ARRAY_LENGTH (regressions); + } - for (i = 0; i < ARRAY_LENGTH (regressions); ++i) + for (i = lo; i < hi; ++i) { const pixel_combination_t *combination = &(regressions[i]); for (j = 1; j < 34; ++j) { - if (!verify (i, combination, j)) + int k, ca; + + ca = combination->mask_format == PIXMAN_null ? 1 : 2; + + for (k = 0; k < ca; ++k) { - result = 1; - break; + if (!verify (i, combination, j, k)) + { + result = 1; + goto next_regression; + } } } + + next_regression: + ; } return result; diff --git a/lib/pixman/test/radial-invalid.c b/lib/pixman/test/radial-invalid.c new file mode 100644 index 000000000..ec85fe320 --- /dev/null +++ b/lib/pixman/test/radial-invalid.c @@ -0,0 +1,54 @@ +#include +#include +#include +#include "utils.h" + +#define WIDTH 100 +#define HEIGHT 100 + +int +main () +{ + pixman_image_t *radial; + pixman_image_t *dest = pixman_image_create_bits ( + PIXMAN_a8r8g8b8, WIDTH, HEIGHT, NULL, -1); + + static const pixman_transform_t xform = + { + { { 0x346f7, 0x0, 0x0 }, + { 0x0, 0x346f7, 0x0 }, + { 0x0, 0x0, 0x10000 } + }, + }; + + static const pixman_gradient_stop_t stops[] = + { + { 0xde61, { 0x4481, 0x96e8, 0x1e6a, 0x29e1 } }, + { 0xfdd5, { 0xfa10, 0xcc26, 0xbc43, 0x1eb7 } }, + { 0xfe1e, { 0xd257, 0x5bac, 0x6fc2, 0xa33b } }, + }; + + static const pixman_point_fixed_t inner = { 0x320000, 0x320000 }; + static const pixman_point_fixed_t outer = { 0x320000, 0x3cb074 }; + + enable_divbyzero_exceptions (); + enable_invalid_exceptions (); + + radial = pixman_image_create_radial_gradient ( + &inner, + &outer, + 0xab074, /* inner radius */ + 0x0, /* outer radius */ + stops, sizeof (stops) / sizeof (stops[0])); + + pixman_image_set_repeat (radial, PIXMAN_REPEAT_REFLECT); + pixman_image_set_transform (radial, &xform); + + pixman_image_composite ( + PIXMAN_OP_OVER, + radial, NULL, dest, + 0, 0, 0, 0, + 0, 0, WIDTH, HEIGHT); + + return 0; +} diff --git a/lib/pixman/test/scaling-test.c b/lib/pixman/test/scaling-test.c index e2f7fa9f4..0ece61110 100644 --- a/lib/pixman/test/scaling-test.c +++ b/lib/pixman/test/scaling-test.c @@ -73,7 +73,7 @@ test_composite (int testnum, pixman_op_t op; pixman_repeat_t repeat = PIXMAN_REPEAT_NONE; pixman_repeat_t mask_repeat = PIXMAN_REPEAT_NONE; - pixman_format_code_t src_fmt, dst_fmt; + pixman_format_code_t src_fmt, mask_fmt, dst_fmt; uint32_t * srcbuf; uint32_t * dstbuf; uint32_t * maskbuf; @@ -145,6 +145,7 @@ test_composite (int testnum, prng_randmemset (dstbuf, dst_stride * dst_height, 0); src_fmt = get_format (src_bpp); + mask_fmt = PIXMAN_a8; dst_fmt = get_format (dst_bpp); if (prng_rand_n (2)) @@ -169,7 +170,7 @@ test_composite (int testnum, src_fmt, src_width, src_height, srcbuf, src_stride); mask_img = pixman_image_create_bits ( - PIXMAN_a8, mask_width, mask_height, maskbuf, mask_stride); + mask_fmt, mask_width, mask_height, maskbuf, mask_stride); dst_img = pixman_image_create_bits ( dst_fmt, dst_width, dst_height, dstbuf, dst_stride); @@ -255,21 +256,6 @@ test_composite (int testnum, else pixman_image_set_filter (mask_img, PIXMAN_FILTER_BILINEAR, NULL, 0); - if (verbose) - { - printf ("src_fmt=%s, dst_fmt=%s\n", - format_name (src_fmt), format_name (dst_fmt)); - printf ("op=%s, scale_x=%d, scale_y=%d, repeat=%d\n", - operator_name (op), scale_x, scale_y, repeat); - printf ("translate_x=%d, translate_y=%d\n", - translate_x, translate_y); - printf ("src_width=%d, src_height=%d, dst_width=%d, dst_height=%d\n", - src_width, src_height, dst_width, dst_height); - printf ("src_x=%d, src_y=%d, dst_x=%d, dst_y=%d\n", - src_x, src_y, dst_x, dst_y); - printf ("w=%d, h=%d\n", w, h); - } - if (prng_rand_n (8) == 0) { pixman_box16_t clip_boxes[2]; @@ -352,10 +338,45 @@ test_composite (int testnum, } if (prng_rand_n (2) == 0) - pixman_image_composite (op, src_img, NULL, dst_img, - src_x, src_y, 0, 0, dst_x, dst_y, w, h); - else - pixman_image_composite (op, src_img, mask_img, dst_img, + { + mask_fmt = PIXMAN_null; + pixman_image_unref (mask_img); + mask_img = NULL; + mask_x = 0; + mask_y = 0; + } + + if (verbose) + { + printf ("op=%s, src_fmt=%s, mask_fmt=%s, dst_fmt=%s\n", + operator_name (op), format_name (src_fmt), + format_name (mask_fmt), format_name (dst_fmt)); + printf ("scale_x=%d, scale_y=%d, repeat=%d, filter=%d\n", + scale_x, scale_y, repeat, src_img->common.filter); + printf ("translate_x=%d, translate_y=%d\n", + translate_x, translate_y); + if (mask_fmt != PIXMAN_null) + { + printf ("mask_scale_x=%d, mask_scale_y=%d, " + "mask_repeat=%d, mask_filter=%d\n", + mask_scale_x, mask_scale_y, mask_repeat, + mask_img->common.filter); + printf ("mask_translate_x=%d, mask_translate_y=%d\n", + mask_translate_x, mask_translate_y); + } + printf ("src_width=%d, src_height=%d, src_x=%d, src_y=%d\n", + src_width, src_height, src_x, src_y); + if (mask_fmt != PIXMAN_null) + { + printf ("mask_width=%d, mask_height=%d, mask_x=%d, mask_y=%d\n", + mask_width, mask_height, mask_x, mask_y); + } + printf ("dst_width=%d, dst_height=%d, dst_x=%d, dst_y=%d\n", + dst_width, dst_height, dst_x, dst_y); + printf ("w=%d, h=%d\n", w, h); + } + + pixman_image_composite (op, src_img, mask_img, dst_img, src_x, src_y, mask_x, mask_y, dst_x, dst_y, w, h); crc32 = compute_crc32_for_image (0, dst_img); @@ -364,7 +385,8 @@ test_composite (int testnum, print_image (dst_img); pixman_image_unref (src_img); - pixman_image_unref (mask_img); + if (mask_img != NULL) + pixman_image_unref (mask_img); pixman_image_unref (dst_img); if (src_stride < 0) diff --git a/lib/pixman/test/solid-test.c b/lib/pixman/test/solid-test.c new file mode 100644 index 000000000..c6ea39770 --- /dev/null +++ b/lib/pixman/test/solid-test.c @@ -0,0 +1,353 @@ +/* + * Copyright © 2015 RISC OS Open Ltd + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. The copyright holders make no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Ben Avison (bavison@riscosopen.org) + * + */ + +#include "utils.h" +#include +#include + +#define WIDTH 32 +#define HEIGHT 32 + +static const pixman_op_t op_list[] = { + PIXMAN_OP_SRC, + PIXMAN_OP_OVER, + PIXMAN_OP_ADD, + PIXMAN_OP_CLEAR, + PIXMAN_OP_SRC, + PIXMAN_OP_DST, + PIXMAN_OP_OVER, + PIXMAN_OP_OVER_REVERSE, + PIXMAN_OP_IN, + PIXMAN_OP_IN_REVERSE, + PIXMAN_OP_OUT, + PIXMAN_OP_OUT_REVERSE, + PIXMAN_OP_ATOP, + PIXMAN_OP_ATOP_REVERSE, + PIXMAN_OP_XOR, + PIXMAN_OP_ADD, + PIXMAN_OP_MULTIPLY, + PIXMAN_OP_SCREEN, + PIXMAN_OP_OVERLAY, + PIXMAN_OP_DARKEN, + PIXMAN_OP_LIGHTEN, + PIXMAN_OP_HARD_LIGHT, + PIXMAN_OP_DIFFERENCE, + PIXMAN_OP_EXCLUSION, +#if 0 /* these use floating point math and are not always bitexact on different platforms */ + PIXMAN_OP_SATURATE, + PIXMAN_OP_DISJOINT_CLEAR, + PIXMAN_OP_DISJOINT_SRC, + PIXMAN_OP_DISJOINT_DST, + PIXMAN_OP_DISJOINT_OVER, + PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_OP_DISJOINT_IN, + PIXMAN_OP_DISJOINT_IN_REVERSE, + PIXMAN_OP_DISJOINT_OUT, + PIXMAN_OP_DISJOINT_OUT_REVERSE, + PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_OP_DISJOINT_XOR, + PIXMAN_OP_CONJOINT_CLEAR, + PIXMAN_OP_CONJOINT_SRC, + PIXMAN_OP_CONJOINT_DST, + PIXMAN_OP_CONJOINT_OVER, + PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_OP_CONJOINT_IN, + PIXMAN_OP_CONJOINT_IN_REVERSE, + PIXMAN_OP_CONJOINT_OUT, + PIXMAN_OP_CONJOINT_OUT_REVERSE, + PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_OP_CONJOINT_XOR, + PIXMAN_OP_COLOR_DODGE, + PIXMAN_OP_COLOR_BURN, + PIXMAN_OP_SOFT_LIGHT, + PIXMAN_OP_HSL_HUE, + PIXMAN_OP_HSL_SATURATION, + PIXMAN_OP_HSL_COLOR, + PIXMAN_OP_HSL_LUMINOSITY, +#endif +}; + +/* The first eight format in the list are by far the most widely + * used formats, so we test those more than the others + */ +#define N_MOST_LIKELY_FORMATS 8 + +static const pixman_format_code_t img_fmt_list[] = { + PIXMAN_a8r8g8b8, + PIXMAN_a8b8g8r8, + PIXMAN_x8r8g8b8, + PIXMAN_x8b8g8r8, + PIXMAN_r5g6b5, + PIXMAN_b5g6r5, + PIXMAN_a8, + PIXMAN_a1, + PIXMAN_r3g3b2, + PIXMAN_b8g8r8a8, + PIXMAN_b8g8r8x8, + PIXMAN_r8g8b8a8, + PIXMAN_r8g8b8x8, + PIXMAN_x14r6g6b6, + PIXMAN_r8g8b8, + PIXMAN_b8g8r8, +#if 0 /* These are going to use floating point in the near future */ + PIXMAN_x2r10g10b10, + PIXMAN_a2r10g10b10, + PIXMAN_x2b10g10r10, + PIXMAN_a2b10g10r10, +#endif + PIXMAN_a1r5g5b5, + PIXMAN_x1r5g5b5, + PIXMAN_a1b5g5r5, + PIXMAN_x1b5g5r5, + PIXMAN_a4r4g4b4, + PIXMAN_x4r4g4b4, + PIXMAN_a4b4g4r4, + PIXMAN_x4b4g4r4, + PIXMAN_r3g3b2, + PIXMAN_b2g3r3, + PIXMAN_a2r2g2b2, + PIXMAN_a2b2g2r2, + PIXMAN_c8, + PIXMAN_g8, + PIXMAN_x4c4, + PIXMAN_x4g4, + PIXMAN_c4, + PIXMAN_g4, + PIXMAN_g1, + PIXMAN_x4a4, + PIXMAN_a4, + PIXMAN_r1g2b1, + PIXMAN_b1g2r1, + PIXMAN_a1r1g1b1, + PIXMAN_a1b1g1r1, + PIXMAN_null +}; + +static const pixman_format_code_t mask_fmt_list[] = { + PIXMAN_a8r8g8b8, + PIXMAN_a8, + PIXMAN_a4, + PIXMAN_a1, + PIXMAN_null +}; + +static pixman_indexed_t rgb_palette[9]; +static pixman_indexed_t y_palette[9]; + +static pixman_format_code_t +random_format (const pixman_format_code_t *allowed_formats) +{ + int n = 0; + + while (allowed_formats[n] != PIXMAN_null) + n++; + + if (n > N_MOST_LIKELY_FORMATS && prng_rand_n (4) != 0) + n = N_MOST_LIKELY_FORMATS; + + return allowed_formats[prng_rand_n (n)]; +} + +static pixman_image_t * +create_multi_pixel_image (const pixman_format_code_t *allowed_formats, + uint32_t *buffer, + pixman_format_code_t *used_fmt) +{ + pixman_format_code_t fmt; + pixman_image_t *img; + int stride; + + fmt = random_format (allowed_formats); + stride = (WIDTH * PIXMAN_FORMAT_BPP (fmt) + 31) / 32 * 4; + img = pixman_image_create_bits (fmt, WIDTH, HEIGHT, buffer, stride); + + if (PIXMAN_FORMAT_TYPE (fmt) == PIXMAN_TYPE_COLOR) + pixman_image_set_indexed (img, &(rgb_palette[PIXMAN_FORMAT_BPP (fmt)])); + else if (PIXMAN_FORMAT_TYPE (fmt) == PIXMAN_TYPE_GRAY) + pixman_image_set_indexed (img, &(y_palette[PIXMAN_FORMAT_BPP (fmt)])); + + prng_randmemset (buffer, WIDTH * HEIGHT * 4, 0); + image_endian_swap (img); + + if (used_fmt) + *used_fmt = fmt; + + return img; +} + +static pixman_image_t * +create_solid_image (const pixman_format_code_t *allowed_formats, + uint32_t *buffer, + pixman_format_code_t *used_fmt) +{ + if (prng_rand_n (2)) + { + /* Use a repeating 1x1 bitmap image for solid */ + pixman_format_code_t fmt; + pixman_image_t *img, *dummy_img; + uint32_t bpp, dummy_buf; + + fmt = random_format (allowed_formats); + bpp = PIXMAN_FORMAT_BPP (fmt); + img = pixman_image_create_bits (fmt, 1, 1, buffer, 4); + pixman_image_set_repeat (img, PIXMAN_REPEAT_NORMAL); + + if (PIXMAN_FORMAT_TYPE (fmt) == PIXMAN_TYPE_COLOR) + pixman_image_set_indexed (img, &(rgb_palette[bpp])); + else if (PIXMAN_FORMAT_TYPE (fmt) == PIXMAN_TYPE_GRAY) + pixman_image_set_indexed (img, &(y_palette[bpp])); + + /* Force the flags to be calculated for image with initial + * bitmap contents of 0 or 2^bpp-1 by plotting from it into a + * separate throwaway image. It is simplest to write all 0s + * or all 1s to the first word irrespective of the colour + * depth even though we actually only care about the first + * pixel since the stride has to be a whole number of words. + */ + *buffer = prng_rand_n (2) ? 0xFFFFFFFFu : 0; + dummy_img = pixman_image_create_bits (PIXMAN_a8r8g8b8, 1, 1, + &dummy_buf, 4); + pixman_image_composite (PIXMAN_OP_SRC, img, NULL, dummy_img, + 0, 0, 0, 0, 0, 0, 1, 1); + pixman_image_unref (dummy_img); + + /* Now set the bitmap contents to a random value */ + prng_randmemset (buffer, 4, 0); + image_endian_swap (img); + + if (used_fmt) + *used_fmt = fmt; + + return img; + } + else + { + /* Use a native solid image */ + pixman_color_t color; + pixman_image_t *img; + + color.alpha = prng_rand_n (UINT16_MAX + 1); + color.red = prng_rand_n (UINT16_MAX + 1); + color.green = prng_rand_n (UINT16_MAX + 1); + color.blue = prng_rand_n (UINT16_MAX + 1); + img = pixman_image_create_solid_fill (&color); + + if (used_fmt) + *used_fmt = PIXMAN_solid; + + return img; + } +} + +static uint32_t +test_solid (int testnum, int verbose) +{ + pixman_op_t op; + uint32_t src_buf[WIDTH * HEIGHT]; + uint32_t dst_buf[WIDTH * HEIGHT]; + uint32_t mask_buf[WIDTH * HEIGHT]; + pixman_image_t *src_img; + pixman_image_t *dst_img; + pixman_image_t *mask_img = NULL; + pixman_format_code_t src_fmt, dst_fmt, mask_fmt = PIXMAN_null; + pixman_bool_t ca = 0; + uint32_t crc32; + + prng_srand (testnum); + + op = op_list[prng_rand_n (ARRAY_LENGTH (op_list))]; + + dst_img = create_multi_pixel_image (img_fmt_list, dst_buf, &dst_fmt); + switch (prng_rand_n (3)) + { + case 0: /* Solid source, no mask */ + src_img = create_solid_image (img_fmt_list, src_buf, &src_fmt); + break; + case 1: /* Solid source, bitmap mask */ + src_img = create_solid_image (img_fmt_list, src_buf, &src_fmt); + mask_img = create_multi_pixel_image (mask_fmt_list, mask_buf, &mask_fmt); + break; + case 2: /* Bitmap image, solid mask */ + src_img = create_multi_pixel_image (img_fmt_list, src_buf, &src_fmt); + mask_img = create_solid_image (mask_fmt_list, mask_buf, &mask_fmt); + break; + default: + abort (); + } + + if (mask_img) + { + ca = prng_rand_n (2); + pixman_image_set_component_alpha (mask_img, ca); + } + + if (verbose) + { + printf ("op=%s\n", operator_name (op)); + printf ("src_fmt=%s, dst_fmt=%s, mask_fmt=%s\n", + format_name (src_fmt), format_name (dst_fmt), + format_name (mask_fmt)); + printf ("src_size=%u, mask_size=%u, component_alpha=%u\n", + src_fmt == PIXMAN_solid ? 1 : src_img->bits.width, + !mask_img || mask_fmt == PIXMAN_solid ? 1 : mask_img->bits.width, + ca); + } + + pixman_image_composite (op, src_img, mask_img, dst_img, + 0, 0, 0, 0, 0, 0, WIDTH, HEIGHT); + + if (verbose) + print_image (dst_img); + + crc32 = compute_crc32_for_image (0, dst_img); + + pixman_image_unref (src_img); + pixman_image_unref (dst_img); + if (mask_img) + pixman_image_unref (mask_img); + + return crc32; +} + +int +main (int argc, const char *argv[]) +{ + int i; + + prng_srand (0); + + for (i = 1; i <= 8; i++) + { + initialize_palette (&(rgb_palette[i]), i, TRUE); + initialize_palette (&(y_palette[i]), i, FALSE); + } + + return fuzzer_test_main ("solid", 500000, + 0xC30FD380, + test_solid, argc, argv); +} diff --git a/lib/pixman/test/thread-test.c b/lib/pixman/test/thread-test.c index 0b07b269d..1c2f0407a 100644 --- a/lib/pixman/test/thread-test.c +++ b/lib/pixman/test/thread-test.c @@ -38,38 +38,11 @@ static const pixman_op_t operators[] = PIXMAN_OP_ATOP_REVERSE, PIXMAN_OP_XOR, PIXMAN_OP_ADD, - PIXMAN_OP_SATURATE, - PIXMAN_OP_DISJOINT_CLEAR, - PIXMAN_OP_DISJOINT_SRC, - PIXMAN_OP_DISJOINT_DST, - PIXMAN_OP_DISJOINT_OVER, - PIXMAN_OP_DISJOINT_OVER_REVERSE, - PIXMAN_OP_DISJOINT_IN, - PIXMAN_OP_DISJOINT_IN_REVERSE, - PIXMAN_OP_DISJOINT_OUT, - PIXMAN_OP_DISJOINT_OUT_REVERSE, - PIXMAN_OP_DISJOINT_ATOP, - PIXMAN_OP_DISJOINT_ATOP_REVERSE, - PIXMAN_OP_DISJOINT_XOR, - PIXMAN_OP_CONJOINT_CLEAR, - PIXMAN_OP_CONJOINT_SRC, - PIXMAN_OP_CONJOINT_DST, - PIXMAN_OP_CONJOINT_OVER, - PIXMAN_OP_CONJOINT_OVER_REVERSE, - PIXMAN_OP_CONJOINT_IN, - PIXMAN_OP_CONJOINT_IN_REVERSE, - PIXMAN_OP_CONJOINT_OUT, - PIXMAN_OP_CONJOINT_OUT_REVERSE, - PIXMAN_OP_CONJOINT_ATOP, - PIXMAN_OP_CONJOINT_ATOP_REVERSE, - PIXMAN_OP_CONJOINT_XOR, PIXMAN_OP_MULTIPLY, PIXMAN_OP_SCREEN, PIXMAN_OP_OVERLAY, PIXMAN_OP_DARKEN, PIXMAN_OP_LIGHTEN, - PIXMAN_OP_COLOR_DODGE, - PIXMAN_OP_COLOR_BURN, PIXMAN_OP_HARD_LIGHT, PIXMAN_OP_DIFFERENCE, PIXMAN_OP_EXCLUSION, @@ -183,7 +156,7 @@ main (void) crc32 = compute_crc32 (0, crc32s, sizeof crc32s); -#define EXPECTED 0xE299B18E +#define EXPECTED 0x82C4D9FB if (crc32 != EXPECTED) { diff --git a/lib/pixman/test/tolerance-test.c b/lib/pixman/test/tolerance-test.c new file mode 100644 index 000000000..320bb7fe0 --- /dev/null +++ b/lib/pixman/test/tolerance-test.c @@ -0,0 +1,360 @@ +#include +#include +#include +#include +#include +#include "utils.h" + +#define MAX_WIDTH 16 +#define MAX_HEIGHT 16 +#define MAX_STRIDE 4 + +static const pixman_format_code_t formats[] = +{ + PIXMAN_a2r10g10b10, + PIXMAN_x2r10g10b10, + PIXMAN_a8r8g8b8, + PIXMAN_a4r4g4b4, + PIXMAN_a2r2g2b2, + PIXMAN_r5g6b5, + PIXMAN_r3g3b2, +}; + +static const pixman_op_t operators[] = +{ + PIXMAN_OP_CLEAR, + PIXMAN_OP_SRC, + PIXMAN_OP_DST, + PIXMAN_OP_OVER, + PIXMAN_OP_OVER_REVERSE, + PIXMAN_OP_IN, + PIXMAN_OP_IN_REVERSE, + PIXMAN_OP_OUT, + PIXMAN_OP_OUT_REVERSE, + PIXMAN_OP_ATOP, + PIXMAN_OP_ATOP_REVERSE, + PIXMAN_OP_XOR, + PIXMAN_OP_ADD, + PIXMAN_OP_SATURATE, + + PIXMAN_OP_DISJOINT_CLEAR, + PIXMAN_OP_DISJOINT_SRC, + PIXMAN_OP_DISJOINT_DST, + PIXMAN_OP_DISJOINT_OVER, + PIXMAN_OP_DISJOINT_OVER_REVERSE, + PIXMAN_OP_DISJOINT_IN, + PIXMAN_OP_DISJOINT_IN_REVERSE, + PIXMAN_OP_DISJOINT_OUT, + PIXMAN_OP_DISJOINT_OUT_REVERSE, + PIXMAN_OP_DISJOINT_ATOP, + PIXMAN_OP_DISJOINT_ATOP_REVERSE, + PIXMAN_OP_DISJOINT_XOR, + + PIXMAN_OP_CONJOINT_CLEAR, + PIXMAN_OP_CONJOINT_SRC, + PIXMAN_OP_CONJOINT_DST, + PIXMAN_OP_CONJOINT_OVER, + PIXMAN_OP_CONJOINT_OVER_REVERSE, + PIXMAN_OP_CONJOINT_IN, + PIXMAN_OP_CONJOINT_IN_REVERSE, + PIXMAN_OP_CONJOINT_OUT, + PIXMAN_OP_CONJOINT_OUT_REVERSE, + PIXMAN_OP_CONJOINT_ATOP, + PIXMAN_OP_CONJOINT_ATOP_REVERSE, + PIXMAN_OP_CONJOINT_XOR, + + PIXMAN_OP_MULTIPLY, + PIXMAN_OP_SCREEN, + PIXMAN_OP_OVERLAY, + PIXMAN_OP_DARKEN, + PIXMAN_OP_LIGHTEN, + PIXMAN_OP_COLOR_DODGE, + PIXMAN_OP_COLOR_BURN, + PIXMAN_OP_HARD_LIGHT, + PIXMAN_OP_SOFT_LIGHT, + PIXMAN_OP_DIFFERENCE, + PIXMAN_OP_EXCLUSION, +}; + +#define RANDOM_ELT(array) \ + (array[prng_rand_n (ARRAY_LENGTH (array))]) + +static void +free_bits (pixman_image_t *image, void *data) +{ + free (image->bits.bits); +} + +static pixman_image_t * +create_image (pixman_image_t **clone) +{ + pixman_format_code_t format = RANDOM_ELT (formats); + pixman_image_t *image; + int width = prng_rand_n (MAX_WIDTH); + int height = prng_rand_n (MAX_HEIGHT); + int stride = ((width * (PIXMAN_FORMAT_BPP (format) / 8)) + 3) & ~3; + uint32_t *bytes = malloc (stride * height); + + prng_randmemset (bytes, stride * height, RANDMEMSET_MORE_00_AND_FF); + + image = pixman_image_create_bits ( + format, width, height, bytes, stride); + + pixman_image_set_destroy_function (image, free_bits, NULL); + + assert (image); + + if (clone) + { + uint32_t *bytes_dup = malloc (stride * height); + + memcpy (bytes_dup, bytes, stride * height); + + *clone = pixman_image_create_bits ( + format, width, height, bytes_dup, stride); + + pixman_image_set_destroy_function (*clone, free_bits, NULL); + } + + return image; +} + +static pixman_bool_t +access (pixman_image_t *image, int x, int y, uint32_t *pixel) +{ + int bytes_per_pixel; + int stride; + uint8_t *location; + + if (x < 0 || x >= image->bits.width || y < 0 || y >= image->bits.height) + return FALSE; + + bytes_per_pixel = PIXMAN_FORMAT_BPP (image->bits.format) / 8; + stride = image->bits.rowstride * 4; + + location = (uint8_t *)image->bits.bits + y * stride + x * bytes_per_pixel; + + if (bytes_per_pixel == 4) + *pixel = *(uint32_t *)location; + else if (bytes_per_pixel == 2) + *pixel = *(uint16_t *)location; + else if (bytes_per_pixel == 1) + *pixel = *(uint8_t *)location; + else + assert (0); + + return TRUE; +} + +static void +get_color (pixel_checker_t *checker, + pixman_image_t *image, + int x, int y, + color_t *color, + uint32_t *pixel) +{ + if (!access (image, x, y, pixel)) + { + color->a = 0.0; + color->r = 0.0; + color->g = 0.0; + color->b = 0.0; + } + else + { + pixel_checker_convert_pixel_to_color ( + checker, *pixel, color); + } +} + +static pixman_bool_t +verify (int test_no, + pixman_op_t op, + pixman_image_t *source, + pixman_image_t *mask, + pixman_image_t *dest, + pixman_image_t *orig_dest, + int x, int y, + int width, int height, + pixman_bool_t component_alpha) +{ + pixel_checker_t dest_checker, src_checker, mask_checker; + int i, j; + + pixel_checker_init (&src_checker, source->bits.format); + pixel_checker_init (&dest_checker, dest->bits.format); + pixel_checker_init (&mask_checker, mask->bits.format); + + assert (dest->bits.format == orig_dest->bits.format); + + for (j = y; j < y + height; ++j) + { + for (i = x; i < x + width; ++i) + { + color_t src_color, mask_color, orig_dest_color, result; + uint32_t dest_pixel, orig_dest_pixel, src_pixel, mask_pixel; + + access (dest, i, j, &dest_pixel); + + get_color (&src_checker, + source, i - x, j - y, + &src_color, &src_pixel); + + get_color (&mask_checker, + mask, i - x, j - y, + &mask_color, &mask_pixel); + + get_color (&dest_checker, + orig_dest, i, j, + &orig_dest_color, &orig_dest_pixel); + + do_composite (op, + &src_color, &mask_color, &orig_dest_color, + &result, component_alpha); + + if (!pixel_checker_check (&dest_checker, dest_pixel, &result)) + { + int a, r, g, b; + + printf ("--------- Test 0x%x failed ---------\n", test_no); + + printf (" operator: %s (%s alpha)\n", operator_name (op), + component_alpha? "component" : "unified"); + printf (" dest_x, dest_y: %d %d\n", x, y); + printf (" width, height: %d %d\n", width, height); + printf (" source: format: %-14s size: %2d x %2d\n", + format_name (source->bits.format), + source->bits.width, source->bits.height); + printf (" mask: format: %-14s size: %2d x %2d\n", + format_name (mask->bits.format), + mask->bits.width, mask->bits.height); + printf (" dest: format: %-14s size: %2d x %2d\n", + format_name (dest->bits.format), + dest->bits.width, dest->bits.height); + printf (" -- Failed pixel: (%d, %d) --\n", i, j); + printf (" source ARGB: %f %f %f %f (pixel: %x)\n", + src_color.a, src_color.r, src_color.g, src_color.b, + src_pixel); + printf (" mask ARGB: %f %f %f %f (pixel: %x)\n", + mask_color.a, mask_color.r, mask_color.g, mask_color.b, + mask_pixel); + printf (" dest ARGB: %f %f %f %f (pixel: %x)\n", + orig_dest_color.a, orig_dest_color.r, orig_dest_color.g, orig_dest_color.b, + orig_dest_pixel); + printf (" expected ARGB: %f %f %f %f\n", + result.a, result.r, result.g, result.b); + + pixel_checker_get_min (&dest_checker, &result, &a, &r, &g, &b); + printf (" min acceptable: %8d %8d %8d %8d\n", a, r, g, b); + + pixel_checker_split_pixel (&dest_checker, dest_pixel, &a, &r, &g, &b); + printf (" got: %8d %8d %8d %8d (pixel: %x)\n", a, r, g, b, dest_pixel); + + pixel_checker_get_max (&dest_checker, &result, &a, &r, &g, &b); + printf (" max acceptable: %8d %8d %8d %8d\n", a, r, g, b); + printf ("\n"); + printf (" { %s,\n", operator_name (op)); + printf (" PIXMAN_%s,\t0x%x,\n", format_name (source->bits.format), src_pixel); + printf (" PIXMAN_%s,\t0x%x,\n", format_name (mask->bits.format), mask_pixel); + printf (" PIXMAN_%s,\t0x%x\n", format_name (dest->bits.format), orig_dest_pixel); + printf (" },\n"); + return FALSE; + } + } + } + + return TRUE; +} + +static pixman_bool_t +do_check (int i) +{ + pixman_image_t *source, *dest, *mask; + pixman_op_t op; + int x, y, width, height; + pixman_image_t *dest_copy; + pixman_bool_t result = TRUE; + pixman_bool_t component_alpha; + + prng_srand (i); + op = RANDOM_ELT (operators); + x = prng_rand_n (MAX_WIDTH); + y = prng_rand_n (MAX_HEIGHT); + width = prng_rand_n (MAX_WIDTH) + 4; + height = prng_rand_n (MAX_HEIGHT) + 4; + + source = create_image (NULL); + mask = create_image (NULL); + dest = create_image (&dest_copy); + + if (x >= dest->bits.width) + x = dest->bits.width / 2; + if (y >= dest->bits.height) + y = dest->bits.height / 2; + if (x + width > dest->bits.width) + width = dest->bits.width - x; + if (y + height > dest->bits.height) + height = dest->bits.height - y; + + component_alpha = prng_rand_n (2); + + pixman_image_set_component_alpha (mask, component_alpha); + + pixman_image_composite32 (op, source, mask, dest, + 0, 0, 0, 0, + x, y, width, height); + + if (!verify (i, op, source, mask, dest, dest_copy, + x, y, width, height, component_alpha)) + { + result = FALSE; + } + + pixman_image_unref (source); + pixman_image_unref (mask); + pixman_image_unref (dest); + pixman_image_unref (dest_copy); + + return result; +} + +#define N_TESTS 10000000 + +int +main (int argc, const char *argv[]) +{ + int i; + int result = 0; + + if (argc == 2) + { + if (strcmp (argv[1], "--forever") == 0) + { + uint32_t n; + + prng_srand (time (0)); + + n = prng_rand(); + + for (;;) + do_check (n++); + } + else + { + do_check (strtol (argv[1], NULL, 0)); + } + } + else + { +#ifdef USE_OPENMP +# pragma omp parallel for default(none) reduction(|:result) +#endif + for (i = 0; i < N_TESTS; ++i) + { + if (!do_check (i)) + result |= 1; + } + } + + return result; +} diff --git a/lib/pixman/test/utils.c b/lib/pixman/test/utils.c index ebe0ccc09..f8e42a5d3 100644 --- a/lib/pixman/test/utils.c +++ b/lib/pixman/test/utils.c @@ -4,6 +4,9 @@ #include #include #include +#include +#include +#include #ifdef HAVE_GETTIMEOFDAY #include @@ -27,6 +30,8 @@ #include #endif +#define ROUND_UP(x, mult) (((x) + (mult) - 1) / (mult) * (mult)) + /* Random number generator state */ @@ -375,7 +380,16 @@ typedef struct int n_bytes; } info_t; -#if defined(HAVE_MPROTECT) && defined(HAVE_GETPAGESIZE) && defined(HAVE_SYS_MMAN_H) && defined(HAVE_MMAP) +#if FENCE_MALLOC_ACTIVE + +unsigned long +fence_get_page_size () +{ + /* You can fake a page size here, if you want to test e.g. 64 kB + * pages on a 4 kB page system. Just put a multiplier below. + */ + return getpagesize (); +} /* This is apparently necessary on at least OS X */ #ifndef MAP_ANONYMOUS @@ -385,7 +399,7 @@ typedef struct void * fence_malloc (int64_t len) { - unsigned long page_size = getpagesize(); + unsigned long page_size = fence_get_page_size (); unsigned long page_mask = page_size - 1; uint32_t n_payload_bytes = (len + page_mask) & ~page_mask; uint32_t n_bytes = @@ -434,7 +448,7 @@ fence_malloc (int64_t len) void fence_free (void *data) { - uint32_t page_size = getpagesize(); + uint32_t page_size = fence_get_page_size (); uint8_t *payload = data; uint8_t *leading_protected = payload - N_LEADING_PROTECTED * page_size; uint8_t *initial_page = leading_protected - page_size; @@ -443,7 +457,98 @@ fence_free (void *data) munmap (info->addr, info->n_bytes); } -#else +static void +fence_image_destroy (pixman_image_t *image, void *data) +{ + fence_free (data); +} + +/* Create an image with fence pages. + * + * Creates an image, where the data area is allocated with fence_malloc (). + * Each row has an additional page in the stride. + * + * min_width is only a minimum width for the image. The width is aligned up + * for the row size to be divisible by both page size and pixel size. + * + * If stride_fence is true, the additional page on each row will be + * armed to cause SIGSEGV or SIGBUS on all accesses. This should catch + * all accesses outside the valid row pixels. + */ +pixman_image_t * +fence_image_create_bits (pixman_format_code_t format, + int min_width, + int height, + pixman_bool_t stride_fence) +{ + unsigned page_size = fence_get_page_size (); + unsigned page_mask = page_size - 1; + unsigned bitspp = PIXMAN_FORMAT_BPP (format); + unsigned bits_boundary; + unsigned row_bits; + int width; /* pixels */ + unsigned stride; /* bytes */ + void *pixels; + pixman_image_t *image; + int i; + + /* must be power of two */ + assert (page_size && (page_size & page_mask) == 0); + + if (bitspp < 1 || min_width < 1 || height < 1) + abort (); + + /* least common multiple between page size * 8 and bitspp */ + bits_boundary = bitspp; + while (! (bits_boundary & 1)) + bits_boundary >>= 1; + bits_boundary *= page_size * 8; + + /* round up to bits_boundary */ + row_bits = ROUND_UP ( (unsigned)min_width * bitspp, bits_boundary); + width = row_bits / bitspp; + + stride = row_bits / 8; + if (stride_fence) + stride += page_size; /* add fence page */ + + if (UINT_MAX / stride < (unsigned)height) + abort (); + + pixels = fence_malloc (stride * (unsigned)height); + if (!pixels) + return NULL; + + if (stride_fence) + { + uint8_t *guard = (uint8_t *)pixels + stride - page_size; + + /* arm row end fence pages */ + for (i = 0; i < height; i++) + { + if (mprotect (guard + i * stride, page_size, PROT_NONE) == -1) + goto out_fail; + } + } + + assert (width >= min_width); + + image = pixman_image_create_bits_no_clear (format, width, height, + pixels, stride); + if (!image) + goto out_fail; + + pixman_image_set_destroy_function (image, fence_image_destroy, pixels); + + return image; + +out_fail: + fence_free (pixels); + + return NULL; +} + +#else /* FENCE_MALLOC_ACTIVE */ void * fence_malloc (int64_t len) @@ -457,7 +562,25 @@ fence_free (void *data) free (data); } -#endif +pixman_image_t * +fence_image_create_bits (pixman_format_code_t format, + int min_width, + int height, + pixman_bool_t stride_fence) +{ + return pixman_image_create_bits (format, min_width, height, NULL, 0); + /* Implicitly allocated storage does not need a destroy function + * to get freed on refcount hitting zero. + */ +} + +unsigned long +fence_get_page_size () +{ + return 0; +} + +#endif /* FENCE_MALLOC_ACTIVE */ uint8_t * make_random_bytes (int n_bytes) @@ -843,9 +966,21 @@ enable_divbyzero_exceptions (void) { #ifdef HAVE_FENV_H #ifdef HAVE_FEENABLEEXCEPT +#ifdef HAVE_FEDIVBYZERO feenableexcept (FE_DIVBYZERO); #endif #endif +#endif +} + +void +enable_invalid_exceptions (void) +{ +#ifdef HAVE_FENV_H +#ifdef HAVE_FEENABLEEXCEPT + feenableexcept (FE_INVALID); +#endif +#endif } void * @@ -937,71 +1072,332 @@ initialize_palette (pixman_indexed_t *palette, uint32_t depth, int is_rgb) } } +struct operator_entry { + pixman_op_t op; + const char *name; + pixman_bool_t is_alias; +}; + +typedef struct operator_entry operator_entry_t; + +static const operator_entry_t op_list[] = +{ +#define ENTRY(op) \ + { PIXMAN_OP_##op, "PIXMAN_OP_" #op, FALSE } +#define ALIAS(op, nam) \ + { PIXMAN_OP_##op, nam, TRUE } + + /* operator_name () will return the first hit in this table, + * so keep the list properly ordered between entries and aliases. + * Aliases are not listed by list_operators (). + */ + + ENTRY (CLEAR), + ENTRY (SRC), + ENTRY (DST), + ENTRY (OVER), + ENTRY (OVER_REVERSE), + ALIAS (OVER_REVERSE, "overrev"), + ENTRY (IN), + ENTRY (IN_REVERSE), + ALIAS (IN_REVERSE, "inrev"), + ENTRY (OUT), + ENTRY (OUT_REVERSE), + ALIAS (OUT_REVERSE, "outrev"), + ENTRY (ATOP), + ENTRY (ATOP_REVERSE), + ALIAS (ATOP_REVERSE, "atoprev"), + ENTRY (XOR), + ENTRY (ADD), + ENTRY (SATURATE), + + ENTRY (DISJOINT_CLEAR), + ENTRY (DISJOINT_SRC), + ENTRY (DISJOINT_DST), + ENTRY (DISJOINT_OVER), + ENTRY (DISJOINT_OVER_REVERSE), + ENTRY (DISJOINT_IN), + ENTRY (DISJOINT_IN_REVERSE), + ENTRY (DISJOINT_OUT), + ENTRY (DISJOINT_OUT_REVERSE), + ENTRY (DISJOINT_ATOP), + ENTRY (DISJOINT_ATOP_REVERSE), + ENTRY (DISJOINT_XOR), + + ENTRY (CONJOINT_CLEAR), + ENTRY (CONJOINT_SRC), + ENTRY (CONJOINT_DST), + ENTRY (CONJOINT_OVER), + ENTRY (CONJOINT_OVER_REVERSE), + ENTRY (CONJOINT_IN), + ENTRY (CONJOINT_IN_REVERSE), + ENTRY (CONJOINT_OUT), + ENTRY (CONJOINT_OUT_REVERSE), + ENTRY (CONJOINT_ATOP), + ENTRY (CONJOINT_ATOP_REVERSE), + ENTRY (CONJOINT_XOR), + + ENTRY (MULTIPLY), + ENTRY (SCREEN), + ENTRY (OVERLAY), + ENTRY (DARKEN), + ENTRY (LIGHTEN), + ENTRY (COLOR_DODGE), + ENTRY (COLOR_BURN), + ENTRY (HARD_LIGHT), + ENTRY (SOFT_LIGHT), + ENTRY (DIFFERENCE), + ENTRY (EXCLUSION), + ENTRY (HSL_HUE), + ENTRY (HSL_SATURATION), + ENTRY (HSL_COLOR), + ENTRY (HSL_LUMINOSITY), + + ALIAS (NONE, "") + +#undef ENTRY +#undef ALIAS +}; + +struct format_entry +{ + pixman_format_code_t format; + const char *name; + pixman_bool_t is_alias; +}; + +typedef struct format_entry format_entry_t; + +static const format_entry_t format_list[] = +{ +#define ENTRY(f) \ + { PIXMAN_##f, #f, FALSE } +#define ALIAS(f, nam) \ + { PIXMAN_##f, nam, TRUE } + + /* format_name () will return the first hit in this table, + * so keep the list properly ordered between entries and aliases. + * Aliases are not listed by list_formats (). + */ + +/* 32bpp formats */ + ENTRY (a8r8g8b8), + ALIAS (a8r8g8b8, "8888"), + ENTRY (x8r8g8b8), + ALIAS (x8r8g8b8, "x888"), + ENTRY (a8b8g8r8), + ENTRY (x8b8g8r8), + ENTRY (b8g8r8a8), + ENTRY (b8g8r8x8), + ENTRY (r8g8b8a8), + ENTRY (r8g8b8x8), + ENTRY (x14r6g6b6), + ENTRY (x2r10g10b10), + ALIAS (x2r10g10b10, "2x10"), + ENTRY (a2r10g10b10), + ALIAS (a2r10g10b10, "2a10"), + ENTRY (x2b10g10r10), + ENTRY (a2b10g10r10), + +/* sRGB formats */ + ENTRY (a8r8g8b8_sRGB), + +/* 24bpp formats */ + ENTRY (r8g8b8), + ALIAS (r8g8b8, "0888"), + ENTRY (b8g8r8), + +/* 16 bpp formats */ + ENTRY (r5g6b5), + ALIAS (r5g6b5, "0565"), + ENTRY (b5g6r5), + + ENTRY (a1r5g5b5), + ALIAS (a1r5g5b5, "1555"), + ENTRY (x1r5g5b5), + ENTRY (a1b5g5r5), + ENTRY (x1b5g5r5), + ENTRY (a4r4g4b4), + ALIAS (a4r4g4b4, "4444"), + ENTRY (x4r4g4b4), + ENTRY (a4b4g4r4), + ENTRY (x4b4g4r4), + +/* 8bpp formats */ + ENTRY (a8), + ALIAS (a8, "8"), + ENTRY (r3g3b2), + ENTRY (b2g3r3), + ENTRY (a2r2g2b2), + ALIAS (a2r2g2b2, "2222"), + ENTRY (a2b2g2r2), + + ALIAS (c8, "x4c4 / c8"), + /* ENTRY (c8), */ + ALIAS (g8, "x4g4 / g8"), + /* ENTRY (g8), */ + + ENTRY (x4a4), + + /* These format codes are identical to c8 and g8, respectively. */ + /* ENTRY (x4c4), */ + /* ENTRY (x4g4), */ + +/* 4 bpp formats */ + ENTRY (a4), + ENTRY (r1g2b1), + ENTRY (b1g2r1), + ENTRY (a1r1g1b1), + ENTRY (a1b1g1r1), + + ALIAS (c4, "c4"), + /* ENTRY (c4), */ + ALIAS (g4, "g4"), + /* ENTRY (g4), */ + +/* 1bpp formats */ + ENTRY (a1), + + ALIAS (g1, "g1"), + /* ENTRY (g1), */ + +/* YUV formats */ + ALIAS (yuy2, "yuy2"), + /* ENTRY (yuy2), */ + ALIAS (yv12, "yv12"), + /* ENTRY (yv12), */ + +/* Fake formats, not in pixman_format_code_t enum */ + ALIAS (null, "null"), + ALIAS (solid, "solid"), + ALIAS (solid, "n"), + ALIAS (pixbuf, "pixbuf"), + ALIAS (rpixbuf, "rpixbuf"), + ALIAS (unknown, "unknown"), + +#undef ENTRY +#undef ALIAS +}; + +pixman_format_code_t +format_from_string (const char *s) +{ + int i; + + for (i = 0; i < ARRAY_LENGTH (format_list); ++i) + { + const format_entry_t *ent = &format_list[i]; + + if (strcasecmp (ent->name, s) == 0) + return ent->format; + } + + return PIXMAN_null; +} + +static void +emit (const char *s, int *n_chars) +{ + *n_chars += printf ("%s,", s); + if (*n_chars > 60) + { + printf ("\n "); + *n_chars = 0; + } + else + { + printf (" "); + (*n_chars)++; + } +} + +void +list_formats (void) +{ + int n_chars; + int i; + + printf ("Formats:\n "); + + n_chars = 0; + for (i = 0; i < ARRAY_LENGTH (format_list); ++i) + { + const format_entry_t *ent = &format_list[i]; + + if (ent->is_alias) + continue; + + emit (ent->name, &n_chars); + } + + printf ("\n\n"); +} + +void +list_operators (void) +{ + char short_name [128] = { 0 }; + int i, n_chars; + + printf ("Operators:\n "); + + n_chars = 0; + for (i = 0; i < ARRAY_LENGTH (op_list); ++i) + { + const operator_entry_t *ent = &op_list[i]; + int j; + + if (ent->is_alias) + continue; + + snprintf (short_name, sizeof (short_name) - 1, "%s", + ent->name + strlen ("PIXMAN_OP_")); + + for (j = 0; short_name[j] != '\0'; ++j) + short_name[j] = tolower (short_name[j]); + + emit (short_name, &n_chars); + } + + printf ("\n\n"); +} + +pixman_op_t +operator_from_string (const char *s) +{ + int i; + + for (i = 0; i < ARRAY_LENGTH (op_list); ++i) + { + const operator_entry_t *ent = &op_list[i]; + + if (ent->is_alias) + { + if (strcasecmp (ent->name, s) == 0) + return ent->op; + } + else + { + if (strcasecmp (ent->name + strlen ("PIXMAN_OP_"), s) == 0) + return ent->op; + } + } + + return PIXMAN_OP_NONE; +} + const char * operator_name (pixman_op_t op) { - switch (op) + int i; + + for (i = 0; i < ARRAY_LENGTH (op_list); ++i) { - case PIXMAN_OP_CLEAR: return "PIXMAN_OP_CLEAR"; - case PIXMAN_OP_SRC: return "PIXMAN_OP_SRC"; - case PIXMAN_OP_DST: return "PIXMAN_OP_DST"; - case PIXMAN_OP_OVER: return "PIXMAN_OP_OVER"; - case PIXMAN_OP_OVER_REVERSE: return "PIXMAN_OP_OVER_REVERSE"; - case PIXMAN_OP_IN: return "PIXMAN_OP_IN"; - case PIXMAN_OP_IN_REVERSE: return "PIXMAN_OP_IN_REVERSE"; - case PIXMAN_OP_OUT: return "PIXMAN_OP_OUT"; - case PIXMAN_OP_OUT_REVERSE: return "PIXMAN_OP_OUT_REVERSE"; - case PIXMAN_OP_ATOP: return "PIXMAN_OP_ATOP"; - case PIXMAN_OP_ATOP_REVERSE: return "PIXMAN_OP_ATOP_REVERSE"; - case PIXMAN_OP_XOR: return "PIXMAN_OP_XOR"; - case PIXMAN_OP_ADD: return "PIXMAN_OP_ADD"; - case PIXMAN_OP_SATURATE: return "PIXMAN_OP_SATURATE"; - - case PIXMAN_OP_DISJOINT_CLEAR: return "PIXMAN_OP_DISJOINT_CLEAR"; - case PIXMAN_OP_DISJOINT_SRC: return "PIXMAN_OP_DISJOINT_SRC"; - case PIXMAN_OP_DISJOINT_DST: return "PIXMAN_OP_DISJOINT_DST"; - case PIXMAN_OP_DISJOINT_OVER: return "PIXMAN_OP_DISJOINT_OVER"; - case PIXMAN_OP_DISJOINT_OVER_REVERSE: return "PIXMAN_OP_DISJOINT_OVER_REVERSE"; - case PIXMAN_OP_DISJOINT_IN: return "PIXMAN_OP_DISJOINT_IN"; - case PIXMAN_OP_DISJOINT_IN_REVERSE: return "PIXMAN_OP_DISJOINT_IN_REVERSE"; - case PIXMAN_OP_DISJOINT_OUT: return "PIXMAN_OP_DISJOINT_OUT"; - case PIXMAN_OP_DISJOINT_OUT_REVERSE: return "PIXMAN_OP_DISJOINT_OUT_REVERSE"; - case PIXMAN_OP_DISJOINT_ATOP: return "PIXMAN_OP_DISJOINT_ATOP"; - case PIXMAN_OP_DISJOINT_ATOP_REVERSE: return "PIXMAN_OP_DISJOINT_ATOP_REVERSE"; - case PIXMAN_OP_DISJOINT_XOR: return "PIXMAN_OP_DISJOINT_XOR"; - - case PIXMAN_OP_CONJOINT_CLEAR: return "PIXMAN_OP_CONJOINT_CLEAR"; - case PIXMAN_OP_CONJOINT_SRC: return "PIXMAN_OP_CONJOINT_SRC"; - case PIXMAN_OP_CONJOINT_DST: return "PIXMAN_OP_CONJOINT_DST"; - case PIXMAN_OP_CONJOINT_OVER: return "PIXMAN_OP_CONJOINT_OVER"; - case PIXMAN_OP_CONJOINT_OVER_REVERSE: return "PIXMAN_OP_CONJOINT_OVER_REVERSE"; - case PIXMAN_OP_CONJOINT_IN: return "PIXMAN_OP_CONJOINT_IN"; - case PIXMAN_OP_CONJOINT_IN_REVERSE: return "PIXMAN_OP_CONJOINT_IN_REVERSE"; - case PIXMAN_OP_CONJOINT_OUT: return "PIXMAN_OP_CONJOINT_OUT"; - case PIXMAN_OP_CONJOINT_OUT_REVERSE: return "PIXMAN_OP_CONJOINT_OUT_REVERSE"; - case PIXMAN_OP_CONJOINT_ATOP: return "PIXMAN_OP_CONJOINT_ATOP"; - case PIXMAN_OP_CONJOINT_ATOP_REVERSE: return "PIXMAN_OP_CONJOINT_ATOP_REVERSE"; - case PIXMAN_OP_CONJOINT_XOR: return "PIXMAN_OP_CONJOINT_XOR"; - - case PIXMAN_OP_MULTIPLY: return "PIXMAN_OP_MULTIPLY"; - case PIXMAN_OP_SCREEN: return "PIXMAN_OP_SCREEN"; - case PIXMAN_OP_OVERLAY: return "PIXMAN_OP_OVERLAY"; - case PIXMAN_OP_DARKEN: return "PIXMAN_OP_DARKEN"; - case PIXMAN_OP_LIGHTEN: return "PIXMAN_OP_LIGHTEN"; - case PIXMAN_OP_COLOR_DODGE: return "PIXMAN_OP_COLOR_DODGE"; - case PIXMAN_OP_COLOR_BURN: return "PIXMAN_OP_COLOR_BURN"; - case PIXMAN_OP_HARD_LIGHT: return "PIXMAN_OP_HARD_LIGHT"; - case PIXMAN_OP_SOFT_LIGHT: return "PIXMAN_OP_SOFT_LIGHT"; - case PIXMAN_OP_DIFFERENCE: return "PIXMAN_OP_DIFFERENCE"; - case PIXMAN_OP_EXCLUSION: return "PIXMAN_OP_EXCLUSION"; - case PIXMAN_OP_HSL_HUE: return "PIXMAN_OP_HSL_HUE"; - case PIXMAN_OP_HSL_SATURATION: return "PIXMAN_OP_HSL_SATURATION"; - case PIXMAN_OP_HSL_COLOR: return "PIXMAN_OP_HSL_COLOR"; - case PIXMAN_OP_HSL_LUMINOSITY: return "PIXMAN_OP_HSL_LUMINOSITY"; - - case PIXMAN_OP_NONE: - return ""; - }; + const operator_entry_t *ent = &op_list[i]; + + if (ent->op == op) + return ent->name; + } return ""; } @@ -1009,95 +1405,164 @@ operator_name (pixman_op_t op) const char * format_name (pixman_format_code_t format) { - switch (format) + int i; + + for (i = 0; i < ARRAY_LENGTH (format_list); ++i) { -/* 32bpp formats */ - case PIXMAN_a8r8g8b8: return "a8r8g8b8"; - case PIXMAN_x8r8g8b8: return "x8r8g8b8"; - case PIXMAN_a8b8g8r8: return "a8b8g8r8"; - case PIXMAN_x8b8g8r8: return "x8b8g8r8"; - case PIXMAN_b8g8r8a8: return "b8g8r8a8"; - case PIXMAN_b8g8r8x8: return "b8g8r8x8"; - case PIXMAN_r8g8b8a8: return "r8g8b8a8"; - case PIXMAN_r8g8b8x8: return "r8g8b8x8"; - case PIXMAN_x14r6g6b6: return "x14r6g6b6"; - case PIXMAN_x2r10g10b10: return "x2r10g10b10"; - case PIXMAN_a2r10g10b10: return "a2r10g10b10"; - case PIXMAN_x2b10g10r10: return "x2b10g10r10"; - case PIXMAN_a2b10g10r10: return "a2b10g10r10"; + const format_entry_t *ent = &format_list[i]; -/* sRGB formats */ - case PIXMAN_a8r8g8b8_sRGB: return "a8r8g8b8_sRGB"; + if (ent->format == format) + return ent->name; + } -/* 24bpp formats */ - case PIXMAN_r8g8b8: return "r8g8b8"; - case PIXMAN_b8g8r8: return "b8g8r8"; - -/* 16bpp formats */ - case PIXMAN_r5g6b5: return "r5g6b5"; - case PIXMAN_b5g6r5: return "b5g6r5"; - - case PIXMAN_a1r5g5b5: return "a1r5g5b5"; - case PIXMAN_x1r5g5b5: return "x1r5g5b5"; - case PIXMAN_a1b5g5r5: return "a1b5g5r5"; - case PIXMAN_x1b5g5r5: return "x1b5g5r5"; - case PIXMAN_a4r4g4b4: return "a4r4g4b4"; - case PIXMAN_x4r4g4b4: return "x4r4g4b4"; - case PIXMAN_a4b4g4r4: return "a4b4g4r4"; - case PIXMAN_x4b4g4r4: return "x4b4g4r4"; + return ""; +}; -/* 8bpp formats */ - case PIXMAN_a8: return "a8"; - case PIXMAN_r3g3b2: return "r3g3b2"; - case PIXMAN_b2g3r3: return "b2g3r3"; - case PIXMAN_a2r2g2b2: return "a2r2g2b2"; - case PIXMAN_a2b2g2r2: return "a2b2g2r2"; - -#if 0 - case PIXMAN_x4c4: return "x4c4"; - case PIXMAN_g8: return "g8"; -#endif - case PIXMAN_c8: return "x4c4 / c8"; - case PIXMAN_x4g4: return "x4g4 / g8"; +#define IS_ZERO(f) (-DBL_MIN < (f) && (f) < DBL_MIN) - case PIXMAN_x4a4: return "x4a4"; +typedef double (* blend_func_t) (double as, double s, double ad, double d); -/* 4bpp formats */ - case PIXMAN_a4: return "a4"; - case PIXMAN_r1g2b1: return "r1g2b1"; - case PIXMAN_b1g2r1: return "b1g2r1"; - case PIXMAN_a1r1g1b1: return "a1r1g1b1"; - case PIXMAN_a1b1g1r1: return "a1b1g1r1"; +static force_inline double +blend_multiply (double sa, double s, double da, double d) +{ + return d * s; +} - case PIXMAN_c4: return "c4"; - case PIXMAN_g4: return "g4"; +static force_inline double +blend_screen (double sa, double s, double da, double d) +{ + return d * sa + s * da - s * d; +} -/* 1bpp formats */ - case PIXMAN_a1: return "a1"; +static force_inline double +blend_overlay (double sa, double s, double da, double d) +{ + if (2 * d < da) + return 2 * s * d; + else + return sa * da - 2 * (da - d) * (sa - s); +} - case PIXMAN_g1: return "g1"; +static force_inline double +blend_darken (double sa, double s, double da, double d) +{ + s = s * da; + d = d * sa; -/* YUV formats */ - case PIXMAN_yuy2: return "yuy2"; - case PIXMAN_yv12: return "yv12"; - }; + if (s > d) + return d; + else + return s; +} - /* Fake formats. - * - * This is separate switch to prevent GCC from complaining - * that the values are not in the pixman_format_code_t enum. - */ - switch ((uint32_t)format) +static force_inline double +blend_lighten (double sa, double s, double da, double d) +{ + s = s * da; + d = d * sa; + + if (s > d) + return s; + else + return d; +} + +static force_inline double +blend_color_dodge (double sa, double s, double da, double d) +{ + if (IS_ZERO (d)) + return 0.0f; + else if (d * sa >= sa * da - s * da) + return sa * da; + else if (IS_ZERO (sa - s)) + return sa * da; + else + return sa * sa * d / (sa - s); +} + +static force_inline double +blend_color_burn (double sa, double s, double da, double d) +{ + if (d >= da) + return sa * da; + else if (sa * (da - d) >= s * da) + return 0.0f; + else if (IS_ZERO (s)) + return 0.0f; + else + return sa * (da - sa * (da - d) / s); +} + +static force_inline double +blend_hard_light (double sa, double s, double da, double d) +{ + if (2 * s < sa) + return 2 * s * d; + else + return sa * da - 2 * (da - d) * (sa - s); +} + +static force_inline double +blend_soft_light (double sa, double s, double da, double d) +{ + if (2 * s <= sa) { - case PIXMAN_null: return "null"; - case PIXMAN_solid: return "solid"; - case PIXMAN_pixbuf: return "pixbuf"; - case PIXMAN_rpixbuf: return "rpixbuf"; - case PIXMAN_unknown: return "unknown"; - }; + if (IS_ZERO (da)) + return d * sa; + else + return d * sa - d * (da - d) * (sa - 2 * s) / da; + } + else + { + if (IS_ZERO (da)) + { + return d * sa; + } + else + { + if (4 * d <= da) + return d * sa + (2 * s - sa) * d * ((16 * d / da - 12) * d / da + 3); + else + return d * sa + (sqrt (d * da) - d) * (2 * s - sa); + } + } +} - return ""; -}; +static force_inline double +blend_difference (double sa, double s, double da, double d) +{ + double dsa = d * sa; + double sda = s * da; + + if (sda < dsa) + return dsa - sda; + else + return sda - dsa; +} + +static force_inline double +blend_exclusion (double sa, double s, double da, double d) +{ + return s * da + d * sa - 2 * d * s; +} + +static double +clamp (double d) +{ + if (d > 1.0) + return 1.0; + else if (d < 0.0) + return 0.0; + else + return d; +} + +static double +blend_channel (double as, double s, double ad, double d, + blend_func_t blend) +{ + return clamp ((1 - ad) * s + (1 - as) * d + blend (as, s, ad, d)); +} static double calc_op (pixman_op_t op, double src, double dst, double srca, double dsta) @@ -1336,6 +1801,21 @@ do_composite (pixman_op_t op, { color_t srcval, srcalpha; + static const blend_func_t blend_funcs[] = + { + blend_multiply, + blend_screen, + blend_overlay, + blend_darken, + blend_lighten, + blend_color_dodge, + blend_color_burn, + blend_hard_light, + blend_soft_light, + blend_difference, + blend_exclusion, + }; + if (mask == NULL) { srcval = *src; @@ -1370,10 +1850,22 @@ do_composite (pixman_op_t op, srcalpha.a = src->a * mask->a; } - result->r = calc_op (op, srcval.r, dst->r, srcalpha.r, dst->a); - result->g = calc_op (op, srcval.g, dst->g, srcalpha.g, dst->a); - result->b = calc_op (op, srcval.b, dst->b, srcalpha.b, dst->a); - result->a = calc_op (op, srcval.a, dst->a, srcalpha.a, dst->a); + if (op >= PIXMAN_OP_MULTIPLY) + { + blend_func_t func = blend_funcs[op - PIXMAN_OP_MULTIPLY]; + + result->a = srcalpha.a + dst->a - srcalpha.a * dst->a; + result->r = blend_channel (srcalpha.r, srcval.r, dst->a, dst->r, func); + result->g = blend_channel (srcalpha.g, srcval.g, dst->a, dst->g, func); + result->b = blend_channel (srcalpha.b, srcval.b, dst->a, dst->b, func); + } + else + { + result->r = calc_op (op, srcval.r, dst->r, srcalpha.r, dst->a); + result->g = calc_op (op, srcval.g, dst->g, srcalpha.g, dst->a); + result->b = calc_op (op, srcval.b, dst->b, srcalpha.b, dst->a); + result->a = calc_op (op, srcval.a, dst->a, srcalpha.a, dst->a); + } } static double @@ -1580,7 +2072,7 @@ get_limits (const pixel_checker_t *checker, double limit, /* The acceptable deviation in units of [0.0, 1.0] */ -#define DEVIATION (0.0064) +#define DEVIATION (0.0128) void pixel_checker_get_max (const pixel_checker_t *checker, color_t *color, diff --git a/lib/pixman/test/utils.h b/lib/pixman/test/utils.h index ebb14d9e4..e299d1d06 100644 --- a/lib/pixman/test/utils.h +++ b/lib/pixman/test/utils.h @@ -86,6 +86,17 @@ is_little_endian (void) void image_endian_swap (pixman_image_t *img); +#if defined (HAVE_MPROTECT) && defined (HAVE_GETPAGESIZE) && \ + defined (HAVE_SYS_MMAN_H) && defined (HAVE_MMAP) +/* fence_malloc and friends have working fence implementation. + * Without this, fence_malloc still allocs but does not catch + * out-of-bounds accesses. + */ +#define FENCE_MALLOC_ACTIVE 1 +#else +#define FENCE_MALLOC_ACTIVE 0 +#endif + /* Allocate memory that is bounded by protected pages, * so that out-of-bounds access will cause segfaults */ @@ -95,6 +106,16 @@ fence_malloc (int64_t len); void fence_free (void *data); +pixman_image_t * +fence_image_create_bits (pixman_format_code_t format, + int min_width, + int height, + pixman_bool_t stride_fence); + +/* Return the page size if FENCE_MALLOC_ACTIVE, or zero otherwise */ +unsigned long +fence_get_page_size (); + /* Generate n_bytes random bytes in fence_malloced memory */ uint8_t * make_random_bytes (int n_bytes); @@ -120,6 +141,7 @@ fail_after (int seconds, const char *msg); /* If possible, enable traps for floating point exceptions */ void enable_divbyzero_exceptions(void); +void enable_invalid_exceptions(void); /* Converts a8r8g8b8 pixels to pixels that * - are not premultiplied, @@ -186,6 +208,18 @@ convert_linear_to_srgb (double component); void initialize_palette (pixman_indexed_t *palette, uint32_t depth, int is_rgb); +pixman_format_code_t +format_from_string (const char *s); + +void +list_formats (void); + +void +list_operators (void); + +pixman_op_t +operator_from_string (const char *s); + const char * operator_name (pixman_op_t op); -- cgit v1.2.3